Friday, 2015-04-24

*** Sukhdev has quit IRC00:01
*** yamada-h has joined #openstack-infra00:01
*** otter768 has joined #openstack-infra00:01
*** sdake has quit IRC00:03
*** apmelton_away is now known as apmelton00:05
*** yamada-h has quit IRC00:06
*** salv-orlando has quit IRC00:07
*** achanda has quit IRC00:07
*** mtanino has quit IRC00:08
*** rkukura has quit IRC00:08
*** salv-orlando has joined #openstack-infra00:08
*** yolanda has joined #openstack-infra00:08
*** tsg_ has quit IRC00:11
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** annegentle has joined #openstack-infra00:15
*** dannywilson has quit IRC00:15
*** ashleighfarnham has joined #openstack-infra00:15
*** otter768 has quit IRC00:19
*** annegentle has quit IRC00:20
*** samueldmq has joined #openstack-infra00:21
*** salv-orlando has quit IRC00:22
*** Krinkle|detached is now known as Krinkle00:23
*** yamamoto has joined #openstack-infra00:23
*** patrickeast has quit IRC00:24
*** ddieterly has joined #openstack-infra00:26
*** ddieterly has quit IRC00:26
*** amitgandhinz has joined #openstack-infra00:27
*** yamamoto has quit IRC00:28
*** gyee has quit IRC00:29
asselinthis nodepool dib command doesn't work in my setup: DEFAULT_QEMU_IMAGE_COMPAT_OPTIONS = "--qemu-img-options 'compat=0.10'"00:32
greghayneswe set that by default now00:33
greghaynesso it shouldnt be needed00:33
asselingreghaynes, used to optional..and now it's not00:33
asselinso my images break00:33
greghayneshow so?00:33
*** tqtran_ has quit IRC00:34
*** yolanda has quit IRC00:35
greghaynesasselin: what qemu-img version is installed?00:35
openstackgerritJeremy Stanley proposed openstack-infra/system-config: Move to HTTPS
cineramapleia2: if you choose to give your seal of approval to i want to hang around when we apply it to translate-dev00:36
asselingreghaynes, this is on ubuntu 1200:37
greghaynesasselin: awesome00:37
greghaynesasselin: so yea, thats a bug00:38
greghaynesWe didnt realize old qemu-img didnt support that argument00:38
asselineasy fix?00:38
greghaynesasselin: Yep, we should check the qemu-img version and set that arg appropriately00:40
*** otter768 has joined #openstack-infra00:40
asselingreghaynes, still not following why it wasn't left optional...?00:43
greghaynesasselin: its something that should be true everywhere and it was proving hard to figure out where the config for it should be00:44
greghaynesand really, theres not a good way for us to describe that in config - its based on the features of the build host not the target00:44
greghaynesexcept as a global config option00:44
asselingreghaynes, so e.g. a config option at the top of nodepool.yaml?00:46
asselingreghaynes, or a DIB_COMPAT option?00:47
*** atan8 has quit IRC00:47
asselinanyway I have to go home. I will think about it and take a closer look tomorrow00:47
*** spzala has joined #openstack-infra00:48
*** nikhil_k is now known as nikhil_k-afk00:48
*** atan8 has joined #openstack-infra00:51
openstackgerritMerged openstack-dev/pbr: Be safe, don't let friends use RawConfigParser
*** esker has quit IRC00:53
*** zhiwei has joined #openstack-infra00:53
*** michchap has quit IRC00:55
*** ddieterly has joined #openstack-infra00:55
krotscheckmordred:, clarkb: Yeah, I'm currently working through the hook approach. It's pretty straightforward to be honest. The trickiest part is making sure your local version of node doesn't suck.00:56
*** ashleighfarnham has quit IRC00:58
*** ddieterly has quit IRC00:58
openstackgerritMerged openstack-infra/system-config: Allow blank lines in node{} with apply test
*** ddieterly has joined #openstack-infra01:00
*** wenlock has joined #openstack-infra01:00
*** jyuso1 has joined #openstack-infra01:03
greghaynesasselin: clarkb
* greghaynes is getting tired of that01:04
*** SumitNaiksatam has quit IRC01:05
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
openstackgerritJeremy Stanley proposed openstack-infra/project-config: There is no Project Bootstrappers-core...
*** spzala has quit IRC01:09
*** tiswanso has joined #openstack-infra01:10
greghaynesif we merge that dib patch then we can just remove the nodepool logic about compat=0.10 entirely01:12
*** tiswanso_ has joined #openstack-infra01:12
*** tiswanso has quit IRC01:14
mordredkrotscheck: sweer01:15
krotscheckI'm having some trouble figuring out the intricacies of extending cmd.Command though01:16
mordredkrotscheck: it's VERY strange01:16
*** tiswanso_ has quit IRC01:17
mordredkrotscheck: there is a sequence where it constructs and "finalizes" args, and then another sequence where things get run01:17
*** tiswanso has joined #openstack-infra01:17
mordredkrotscheck: it's one of the most obtuse things ever, and you essentially have to read down into distutils and also the setuptools layer to grok it01:18
krotscheckmordred: I'm guessing somewhere in the middle there it reads options from other places and assigns them to the instance. Feels... Java POJO-Mapping-ish01:18
*** baoli has quit IRC01:18
krotscheckmordred: Like, you have to tell it that there's a property that can be assigned there before it'll actually assign that property.01:18
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
openstackgerritgreghaynes proposed openstack-infra/nodepool: Stop setting compat=0.10 for qemu-img
mordredkrotscheck: yah. very gross01:19
openstackgerritJeremy Stanley proposed openstack-infra/system-config: Add a centos-6 node label
*** bknudson has joined #openstack-infra01:21
*** ChuckC has quit IRC01:21
*** sdake has joined #openstack-infra01:22
*** atan8 has quit IRC01:23
*** sdake_ has quit IRC01:26
*** michchap has joined #openstack-infra01:26
*** flaper87 has quit IRC01:27
*** emagana has quit IRC01:27
*** flaper87 has joined #openstack-infra01:28
*** liusheng has joined #openstack-infra01:29
*** atan8 has joined #openstack-infra01:29
*** macjack has left #openstack-infra01:32
*** flaper87 has quit IRC01:33
*** zz_dimtruck is now known as dimtruck01:35
*** salv-orlando has joined #openstack-infra01:35
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** nelsnelson has joined #openstack-infra01:38
*** flaper87 has joined #openstack-infra01:39
*** kaisers has quit IRC01:40
*** nelsnelson has quit IRC01:41
*** kaisers has joined #openstack-infra01:42
*** nelsnelson has joined #openstack-infra01:43
*** salv-orlando has quit IRC01:45
*** atan8 has quit IRC01:48
Kennan2hi, :clarkb01:49
*** amitgandhinz has quit IRC01:49
Kennan2there ?01:49
*** amitgandhinz has joined #openstack-infra01:49
*** ChuckC has joined #openstack-infra01:49
*** sdake_ has joined #openstack-infra01:50
Kennan2a question about how to enable our project (master branch) devstack gate run against other projects kilo branch ?01:50
*** shardy_ has joined #openstack-infra01:50
Kennan2since our project not branched now01:50
Kennan2I found the devstack gate job run against other projects all with master01:50
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
Kennan2I am not sure how to make it run against with kilo branch ?01:50
Kennan2did I need configure something in devstack ?01:51
Kennan2:clarkb are you here ? :)01:51
*** shardy has quit IRC01:51
*** sdake has quit IRC01:53
*** ChuckC has quit IRC01:53
*** ChuckC has joined #openstack-infra01:54
*** garyk has quit IRC01:54
*** amitgandhinz has quit IRC01:54
Kennan2:lifeless there ?01:54
*** garyk has joined #openstack-infra01:55
*** Kennan2 is now known as Kennan01:55
*** shardy_ has quit IRC01:55
*** shardy has joined #openstack-infra01:55
lifelessKennan: yes?01:56
KennanHI :lifeless01:56
Kennanour project now not branched01:56
Kennanstill master01:56
*** asettle is now known as asettle-luncheon01:57
Kennanbut other projects in openstack (like nova etc, branched with kilo)01:57
Kennanwe want our devstack gate job still run against other projects kilo01:57
Kennando you know how to do that ?01:57
*** apmelton is now known as apmelton_away01:57
Kennansince if run again master, some requirements version can cause issue01:57
lifelessI'm not sure01:58
Kennando you know what make devstack gate job run against kilo or master branch ?01:58
Kennanis it setup by jenkins ?01:58
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** Sam_Wan has quit IRC02:01
lifelessKennan: its all controlled by jjb02:01
*** harlowja is now known as harlowja_away02:02
*** xianghuihui has joined #openstack-infra02:02
*** baoli has joined #openstack-infra02:03
krotscheckWaaaait a sec.02:03
krotscheckThe only way to add a command into the setuptools build chain is.... to monkeypatch the command?02:04
*** xianghuihui has left #openstack-infra02:06
lifelesskrotscheck: you reallllly don't want to stare down that path02:06
krotschecklifeless: Iiiii'm starting to get that.02:07
*** mwagner_lap has quit IRC02:07
*** michchap_ has joined #openstack-infra02:08
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
* krotscheck walks away slowly02:10
*** baoli has quit IRC02:10
*** michchap has quit IRC02:12
*** baoli has joined #openstack-infra02:12
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** rfolco has quit IRC02:15
*** salv-orlando has joined #openstack-infra02:15
*** ivar-lazzaro has quit IRC02:16
*** baoli has quit IRC02:19
*** tkelsey has joined #openstack-infra02:19
*** tqtran has joined #openstack-infra02:19
*** salv-orlando has quit IRC02:20
*** mwagner_lap has joined #openstack-infra02:21
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** xyang1 has quit IRC02:23
*** tkelsey has quit IRC02:23
*** dannywilson has joined #openstack-infra02:23
*** dannywilson has quit IRC02:23
*** tqtran has quit IRC02:24
*** dannywilson has joined #openstack-infra02:24
stevemarquestion - if we wanted to rename the project `python-openstackclient` to just `openstackclient` how much of a PITA is that?02:26
stevemarwould it just be the same process as a new library?02:26
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: puppet: use a single image for CI
lifelessstevemar: essentially yes02:29
openstackgerritTimo Tijhof proposed openstack-infra/zuul: Status: Use vertical-align for progress bar instead of margin
openstackgerritTimo Tijhof proposed openstack-infra/zuul: Status: Increase width of change-progress-row-left
lifelessstevemar: though, the impact on requirements in many projects may be a bit tedious02:29
lifelessstevemar: (doing point releases to stable branches with a different release name...)02:30
stevemarlifeless, shouldn't be *that* many projects depending on us, just a few of the other client libs and devstack02:31
*** SumitNaiksatam has joined #openstack-infra02:31
lifelessstevemar: its not the # :)02:31
lifelessstevemar: its the location02:31
*** jyuso1 has quit IRC02:37
*** marun has joined #openstack-infra02:37
*** achanda has joined #openstack-infra02:37
*** wenlock has quit IRC02:39
*** unicell has quit IRC02:39
*** baoli has joined #openstack-infra02:40
*** baoli_ has joined #openstack-infra02:42
*** baoli has quit IRC02:42
dtroyerlifeless: All of those would be packages, pypi and the like, though, right?  As long as the last release remained they should work until updated02:43
lifelessdtroyer: sure, the thing I raised was doing point releases02:43
lifelesswe've made stable branches of everything now AIUI02:43
lifelessso if you want to do a point release of python-openstackclient02:43
dtroyerwe'd have to fake it in pypi02:43
lifelessyou'd need the infra tooling for openstackclient to know how to publish to openstackclient for master and python-openstackclient for kilo02:44
lifelessor bypass it all02:44
stevemaroh i didn't think about that lifeless02:44
ianwTIL : how broken bash errexit is02:45
dtroyerin an emergency I'd want to do that, otherwise make the changes before needing a point release02:45
* dtroyer crosses fingers02:45
lifelessdtroyer: 18 months for kilo's lifetime right? :)02:45
stevemardtroyer, i'm going to push a renamed version to my private repo02:46
stevemarthough i have no idea how to maintain git history02:46
dtroyerlifeless: I think, or is it slightly longer? 21 maybe?02:46
dtroyerstevemar: that's doable but I don't know how infra does it02:47
dtroyerthey've got it figured02:47
stevemarwell tests pass, and pep802:48
stevemarall my renaming worked02:48
*** wuhg has quit IRC02:49
dtroyerwhat if we released in parallel for a time?  Both names, same versions02:49
stevemarthat sounds like a maintenance nightmare02:50
dtroyerif it was scripted, maybe not…only one repo02:50
dtroyerand only the code, document everything with the new name02:50
dtroyerjust thinking out loud02:51
dtroyerdamn, ruby installs s..l..o..w on a RasPi02:51
*** bhuvan has quit IRC02:52
stevemardtroyer, slower than osc runs?02:54
*** ayoung has quit IRC02:56
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** tiswanso has quit IRC02:58
*** sdake has joined #openstack-infra03:00
*** marun has quit IRC03:01
openstackgerritTimo Tijhof proposed openstack-infra/zuul: Add cache-hardlink argument to cloner
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** yamada-h has joined #openstack-infra03:03
*** sdake_ has quit IRC03:04
*** sdake_ has joined #openstack-infra03:04
dtroyerianw: ugh, I think it was there already…03:05
stevemardtroyer, do you use run_tests?03:06
stevemaror just tox?03:06
dtroyerin osc?  just tox03:06
stevemari'm going to propose we remove it03:06
stevemardtroyer, what about the stuff in tools? (install_venv and such?)03:07
*** yamada-h has quit IRC03:07
dtroyerunsure if that's still needed.  are they in the cookiecuter output?03:08
*** sdake has quit IRC03:08
* stevemar shrugs03:08
dtroyerdoesn't look like it03:08
stevemarha, OSC has 998 commits03:08
stevemarthats pretty cool03:09
stevemarwe should rename it at 1000 :P03:09
dtroyercommit 1000 will be a giant replacement with roscoe, my go lcient ;)03:10
dtroyeror client even03:10
stevemardtroyer, ?03:11
dtroyerthat's the early version03:11
dtroyerfrom when I threatend to use C and libcurl03:12
stevemarshould have stuck with your threat03:13
stevemardtroyer, bump:
* stevemar just realized that we've been chatting in -infra, my bad03:14
dtroyerstevemar: is more recent.  I haven't pushed since I started re-working stackforge/golang-client last weekend03:14
dtroyerwanna see something really funny?03:14
dtroyerOSC's now vanished-from-the-scene father:
ianwdtroyer: it sort of is ... but did you know that "function || echo 'i failed' " will disable errexit in "function" ?  it seems the path for every developer to knowing this is via about an hour of debugging03:15
ianwto find out that infact that is the POSIX way03:16
dtroyerianw: I thought that was only in pipelines, not boolean operators03:16
dtroyerdang, we need to dump that shortcut if so03:16
ianw"... part of a command executed in a && or || list except the command following the final && or || ..."03:18
ianwdon't google it, it will just make you angry looking at other people who are angry and bash developers who are also angry because they follow posix on it and, well, everyone is angry :)03:19
*** samueldmq has quit IRC03:20
dtroyerI don't need angry tonight, thanks for the warning03:21
dtroyeryay POSIX03:21
*** yamahata has joined #openstack-infra03:23
*** yamamoto has joined #openstack-infra03:32
*** shashankhegde has quit IRC03:32
*** Longgeek has joined #openstack-infra03:33
*** ddieterly has quit IRC03:40
*** xianghuihui has joined #openstack-infra03:40
*** xianghuihui has quit IRC03:40
*** xianghuihui has joined #openstack-infra03:41
*** xianghuihui has quit IRC03:41
*** esker has joined #openstack-infra03:42
*** achanda has quit IRC03:44
openstackgerritTimo Tijhof proposed openstack-infra/zuul: Add cache-no-hardlinks argument to zuul-cloner
openstackgerritTimo Tijhof proposed openstack-infra/zuul: Cloner: Implement cache-no-hardlinks argument
*** esker has quit IRC03:47
*** Longgeek has quit IRC03:51
*** tsg has joined #openstack-infra03:52
*** otter768 has quit IRC03:54
*** asettle-luncheon is now known as asettle03:54
*** Krinkle is now known as Krinkle|detached03:59
*** Longgeek has joined #openstack-infra04:00
*** wenlock has joined #openstack-infra04:00
*** yamada-h has joined #openstack-infra04:04
*** baoli_ has quit IRC04:06
*** yamada-h has quit IRC04:08
*** maishsk_ has joined #openstack-infra04:14
maishsk_pleia2: Are you around?04:14
*** salv-orlando has joined #openstack-infra04:18
*** yamamoto has quit IRC04:24
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** esker has joined #openstack-infra04:25
*** salv-orlando has quit IRC04:25
*** subscope_ has joined #openstack-infra04:30
*** atan8 has joined #openstack-infra04:30
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** achanda has joined #openstack-infra04:35
*** kaisers has quit IRC04:37
*** yamamoto has joined #openstack-infra04:38
*** Longgeek has quit IRC04:42
*** unicell has joined #openstack-infra04:42
*** emagana has joined #openstack-infra04:45
*** mrmartin has joined #openstack-infra04:45
maishsk_tristanC: pleia2: Ping!!04:46
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** markvoelker_ has quit IRC04:51
*** Longgeek has joined #openstack-infra04:52
*** wuhg has joined #openstack-infra04:52
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
*** dhritishikhar_ has quit IRC04:53
*** tqtran has joined #openstack-infra04:53
*** kaisers has joined #openstack-infra04:54
*** nelsnelson has quit IRC04:55
*** dhritishikhar has joined #openstack-infra04:55
*** salv-orlando has joined #openstack-infra04:58
*** woodster_ has quit IRC05:00
*** macjack has joined #openstack-infra05:02
*** dannywilson has quit IRC05:02
*** salv-orlando has quit IRC05:05
*** AJaeger has joined #openstack-infra05:08
*** sarob has joined #openstack-infra05:09
sarobi need to get merged05:10
sarobanyone up for reviews?05:10
*** ddieterly has joined #openstack-infra05:11
morganfainbergsarob: you might want to go with experimental instead of removed, you could then "check experimental" and see if it passes. (As a thought)05:12
morganfainbergBut it doesn't run unless specifically asked.05:12
sarobah, right05:12
sarobits more about im not ready to run tests yet05:13
morganfainbergRight. Makes sense.05:13
saroband the ones applied to this repo would be the wrong ones05:13
sarobwhen we do start05:13
morganfainbergAh that makes more sense then to remove.05:13
sarobthanks for the advice though05:14
*** ddieterly has quit IRC05:16
*** tnovacik has joined #openstack-infra05:16
*** garyk has quit IRC05:16
*** maishsk_ has quit IRC05:17
*** deepakcs has joined #openstack-infra05:17
*** salv-orlando has joined #openstack-infra05:17
*** maishsk has joined #openstack-infra05:17
*** sks has joined #openstack-infra05:18
*** salv-orlando has quit IRC05:21
openstackgerritIan Wienand proposed openstack-infra/devstack-gate: [WIP] Add -e to gate wrap script
openstackgerritAndreas Jaeger proposed openstack-infra/project-config: Add Elastic Recheck to
AJaegersarob: is that only for this repo or also for the other akanada ones?05:25
sarobonly for this repo05:25
AJaegerjhesketh, SergeyLukjanov: Could you review 176948, please? We need one more +2...05:26
AJaegersarob: I pinged the other two core reviewers that might be awake ;) ^05:26
*** atan8 has quit IRC05:26
*** wenlock has quit IRC05:27
jheskethAJaeger: looking05:28
*** BharatK has joined #openstack-infra05:29
openstackgerritAndreas Jaeger proposed openstack/requirements: Remove doc repositories from projects.txt
AJaegerthanks, jhesketh !05:29
jheskethAJaeger, sarob: lgtm, approved05:29
sarobthanks people05:30
sarobone down, 500 to go05:30
*** maishsk has quit IRC05:30
*** afazekas has quit IRC05:31
AJaegersarob: wait 30 mins to have it deployed...05:32
sarobits late here, i will in the wee hours05:33
*** maishsk has joined #openstack-infra05:35
openstackgerritDi Xu proposed openstack-infra/jenkins-job-builder: Support deep merge of defaults
*** tsg has quit IRC05:36
*** yolanda has joined #openstack-infra05:37
*** dimtruck is now known as zz_dimtruck05:40
*** hdd has joined #openstack-infra05:40
openstackgerritMerged openstack-infra/project-config: Update to akanda project
*** e0ne has joined #openstack-infra05:43
*** e0ne is now known as e0ne_05:43
*** e0ne_ is now known as e0ne05:43
openstackgerritMehdi Abaakouk (sileht) proposed openstack/requirements: Bump kombu version and add amqp version
openstackgerritMerged openstack-infra/project-config: extends gerrit acls to allow creation of stable branches
*** e0ne has quit IRC05:45
openstackgerrityolanda.robla proposed openstack-infra/system-config: Move manage_projects logic to puppet-jeepyb
openstackgerritMerged openstack-infra/project-config: Add publishing, check-requirement jobs to os-brick
*** jyuso1 has joined #openstack-infra05:49
*** afazekas has joined #openstack-infra05:49
*** mrunge has joined #openstack-infra05:51
openstackgerritDi Xu proposed openstack-infra/jenkins-job-builder: Support deep merge of defaults
*** sarob has quit IRC05:54
*** otter768 has joined #openstack-infra05:55
*** vlaza has joined #openstack-infra05:57
openstackgerritMerged openstack-infra/project-config: Enable branch creation for networking-cisco repo
*** sarob has joined #openstack-infra05:59
*** scheuran has joined #openstack-infra06:00
*** otter768 has quit IRC06:00
*** david-ly_ has joined #openstack-infra06:01
*** david-lyle has quit IRC06:03
*** subscope_ has quit IRC06:11
*** ddieterly has joined #openstack-infra06:12
*** tqtran has quit IRC06:14
*** ddieterly has quit IRC06:16
*** dhritishikhar has quit IRC06:16
*** dhritishikhar has joined #openstack-infra06:16
*** unicell has quit IRC06:21
*** unicell has joined #openstack-infra06:23
openstackgerritAndreas Jaeger proposed openstack-infra/project-config: Fix translation compress for manuals
*** maishsk has quit IRC06:29
*** stevemar has quit IRC06:32
AJaegerjhesketh, SergeyLukjanov: If you're reviewing today, please consider 177092 ^06:34
jheskethAJaeger: lgtm but I can't test at the moment so only +106:37
AJaegerjhesketh: I tested locally - I should have tested initially on *all* repos...06:38
AJaegerjhesketh: thansk06:38
*** chlong has quit IRC06:39
*** mpaolino has joined #openstack-infra06:42
* AJaeger just noticed debugging code in it - and will double check now06:42
openstackgerrityolanda.robla proposed openstack-infra/system-config: Move manage_projects logic to puppet-jeepyb
*** salv-orlando has joined #openstack-infra06:45
*** Ala has joined #openstack-infra06:50
*** armax has quit IRC06:51
openstackgerritAndreas Jaeger proposed openstack-infra/project-config: Fix translation compress for manuals
*** sarob has quit IRC06:53
*** sarob_ has joined #openstack-infra06:54
*** salv-orlando has quit IRC06:55
*** sarob_ has quit IRC06:58
*** dtantsur|afk is now known as dtantsur07:02
*** luqas has joined #openstack-infra07:06
*** luqas has quit IRC07:06
*** yamada-h has joined #openstack-infra07:06
*** luqas has joined #openstack-infra07:06
*** coolsvap|afk is now known as coolsvap07:08
*** coolsvap is now known as coolsvap|afk07:09
*** tkelsey has joined #openstack-infra07:09
*** yamada-h has quit IRC07:10
*** salv-orlando has joined #openstack-infra07:12
openstackgerritDavid Pursehouse proposed openstack-infra/lodgeit: Add deploye WSGI with proper database path
*** ddieterly has joined #openstack-infra07:13
*** tkelsey has quit IRC07:13
*** salv-orlando has quit IRC07:14
*** ddieterly has quit IRC07:17
*** sdake_ has quit IRC07:20
*** luqas has quit IRC07:22
*** tqtran has joined #openstack-infra07:23
*** achanda has quit IRC07:24
*** dizquierdo has joined #openstack-infra07:27
*** tqtran has quit IRC07:27
openstackgerrityolanda.robla proposed openstack-infra/system-config: Move manage_projects logic to puppet-jeepyb
*** wuhg has quit IRC07:29
*** Hal has joined #openstack-infra07:29
*** Hal is now known as Guest7502607:30
openstackgerritDavid Pursehouse proposed openstack-infra/lodgeit: Update i18n files
openstackgerritDavid Pursehouse proposed openstack-infra/lodgeit: Remove out-of-date paragraph from the 'not found' page
*** arxcruz has joined #openstack-infra07:30
*** achanda has joined #openstack-infra07:36
*** [HeOS] has quit IRC07:39
*** salv-orlando has joined #openstack-infra07:40
*** tkelsey has joined #openstack-infra07:42
*** jistr has joined #openstack-infra07:44
*** scheuran has quit IRC07:45
*** tkelsey has quit IRC07:46
*** e0ne has joined #openstack-infra07:47
*** scheuran has joined #openstack-infra07:47
*** scheuran has quit IRC07:48
*** e0ne has quit IRC07:49
*** kaisers1 has joined #openstack-infra07:50
*** yamamoto has quit IRC07:50
*** yolanda has quit IRC07:51
*** zz_johnthetubagu is now known as johnthetubaguy07:51
*** jlanoux has joined #openstack-infra07:51
*** kaisers2 has joined #openstack-infra07:52
*** e0ne has joined #openstack-infra07:53
*** hichihara has quit IRC07:54
*** otter768 has joined #openstack-infra07:56
*** tonytan4ever has quit IRC07:57
*** achanda has quit IRC07:57
*** jcoufal has joined #openstack-infra07:57
*** kaisers2 has quit IRC07:59
*** dtantsur is now known as dtantsur|brb08:00
*** otter768 has quit IRC08:01
*** e0ne has quit IRC08:05
*** ajo has joined #openstack-infra08:06
*** e0ne has joined #openstack-infra08:09
*** ajo_ has joined #openstack-infra08:09
*** ajo has quit IRC08:09
*** jistr is now known as jistr|mtg08:09
*** dizquierdo has quit IRC08:09
*** e0ne has quit IRC08:11
*** hashar has joined #openstack-infra08:11
openstackgerritSalvatore Orlando proposed openstack-infra/elastic-recheck: Add a query for bug 1285323
openstackbug 1285323 in grenade "Services fail to shut down on the old side of Grenade" [Undecided,In progress] - Assigned to Sean Dague (sdague)08:12
*** ddieterly has joined #openstack-infra08:13
*** fhubik has joined #openstack-infra08:14
*** ssam2 has joined #openstack-infra08:16
*** yamamoto has joined #openstack-infra08:18
*** hyakuhei has joined #openstack-infra08:18
*** ddieterly has quit IRC08:18
*** devvesa has joined #openstack-infra08:18
*** jyuso1 has quit IRC08:19
*** hyakuhei has quit IRC08:19
*** hyakuhei has joined #openstack-infra08:23
*** ihrachyshka has joined #openstack-infra08:23
*** _nadya_ has joined #openstack-infra08:24
*** gilliard is now known as gillliard08:24
*** mpaolino has quit IRC08:26
*** andreykurilin__ has joined #openstack-infra08:27
openstackgerritDavide Guerri proposed openstack-infra/shade: WiP: Add Neutron/Nova Floating IP support
openstackgerritJulien Danjou proposed openstack-infra/project-config: Run Gnocchi devstack tests for stable/1.0 against Kilo
*** MaxV has joined #openstack-infra08:29
openstackgerritJulien Danjou proposed openstack-infra/project-config: Publish Gnocchi to PyPI
*** maishsk has joined #openstack-infra08:34
*** derekh has joined #openstack-infra08:35
*** hyakuhei has quit IRC08:36
*** weshay has joined #openstack-infra08:40
*** _nadya_ has quit IRC08:40
*** salv-orlando has quit IRC08:43
openstackgerritMerged openstack/requirements: Bump kombu version and add amqp version
*** [HeOS] has joined #openstack-infra08:47
openstackgerritVictor Stinner proposed openstack-infra/jenkins-job-builder: Fix Builder.update_job() on Python 3
*** SumitNaiksatam has quit IRC08:51
*** ajo_ has quit IRC08:51
*** ajo has joined #openstack-infra08:59
*** jkt has quit IRC09:00
*** jkt has joined #openstack-infra09:01
openstackgerritDawid Malinowski proposed openstack-infra/jenkins-job-builder: Add support for "Display Upstream Changes Plugin".
*** _nadya_ has joined #openstack-infra09:05
*** fhubik is now known as fhubik_afk09:07
*** cody-somerville has joined #openstack-infra09:07
*** yamada-h has joined #openstack-infra09:08
*** e0ne has joined #openstack-infra09:10
*** ajo has quit IRC09:10
openstackgerritDavide Guerri proposed openstack-infra/shade: WiP: Add Neutron/Nova Floating IP support
*** yamada-h has quit IRC09:12
*** ddieterly has joined #openstack-infra09:14
*** e0ne is now known as e0ne_09:16
*** ajo has joined #openstack-infra09:16
*** yolanda has joined #openstack-infra09:18
*** ddieterly has quit IRC09:19
*** pelix has joined #openstack-infra09:20
*** e0ne_ has quit IRC09:21
*** jyuso1 has joined #openstack-infra09:28
*** jistr|mtg is now known as jistr09:32
*** teran has quit IRC09:33
*** jerryz has joined #openstack-infra09:33
*** teran has joined #openstack-infra09:33
*** teran has quit IRC09:33
*** e0ne has joined #openstack-infra09:35
*** salv-orlando has joined #openstack-infra09:35
openstackgerritIhar Hrachyshka proposed openstack-dev/hacking: Add support for flake8 off_by_default for optional checks
*** sdake has joined #openstack-infra09:37
*** fhubik_afk is now known as fhubik09:38
*** sdake_ has joined #openstack-infra09:39
*** dizquierdo has joined #openstack-infra09:40
openstackgerritMehdi Abaakouk (sileht) proposed openstack/requirements: Bump kombu version and add amqp version
*** sdake has quit IRC09:43
*** hyakuhei has joined #openstack-infra09:44
*** Longgeek_ has joined #openstack-infra09:52
*** Longgeek has quit IRC09:52
Kennan:Ajaeger or :fungi there09:55
AJaegerHi Kennan !09:55
KennanHi :Ajaeger09:55
KennanI have one question about: as our project not branched yet09:55
AJaegerwhich repo?09:56
Kennanwe not have kilo branch as others, magnum09:56
Kennanwe want our code gate job still run aginast with stable/kilo09:56
Kennando you know how to make that happen ?09:56
*** yamamoto has quit IRC09:56
*** cdent has joined #openstack-infra09:57
AJaegerthe gate jobs are setup that they will run with any branch (unless you explictely disable it running somewhere)09:57
AJaegerSo, once you have a stable/kilo branch, it should work from the infra side just fine09:57
*** otter768 has joined #openstack-infra09:57
AJaegerBtw. right now you do not have permssions in project-config file gerrit/acls/openstack/magnum.config setup for creation of branches09:58
AJaegerKennan: please read this one here in general for release management:
Kennan:Ajaeger so you means,  if we only have master branch, it would definitely run against with devstack (master branch for all projects, nova etc.)10:00
*** yamamoto has joined #openstack-infra10:00
Kennanwe can not make it run stable/kilo ?10:00
Kennanfor example,  the Julien Danjou proposed to run their branch ?10:02
Kennanis this a working way to solve such issue ?10:02
*** otter768 has quit IRC10:03
AJaegerKennan: that's a question you better discuss with others, can't help here. sdague, fungi, or clarkb should be around in 3hours to discuss10:04
Kennanok. sure. thanks guy.10:04
*** Guest75026 has quit IRC10:04
*** flaper87 has quit IRC10:05
*** flaper87 has joined #openstack-infra10:05
*** markvoelker has joined #openstack-infra10:06
*** pc_m has joined #openstack-infra10:09
*** pc_m has quit IRC10:10
*** fhubik has quit IRC10:11
*** markvoelker has quit IRC10:11
*** fhubik has joined #openstack-infra10:11
*** dhritishikhar has quit IRC10:13
*** dhritishikhar has joined #openstack-infra10:15
*** ddieterly has joined #openstack-infra10:15
*** hyakuhei has quit IRC10:15
*** fhubik has quit IRC10:16
*** fhubik_afk has joined #openstack-infra10:16
*** fhubik_afk is now known as fhubik10:16
*** pc_m has joined #openstack-infra10:18
*** ddieterly has quit IRC10:19
*** Bsony has joined #openstack-infra10:20
*** Longgeek_ has quit IRC10:22
*** Longgeek has joined #openstack-infra10:23
*** maishsk has quit IRC10:24
*** samueldmq has joined #openstack-infra10:25
*** andreykurilin__ has quit IRC10:26
*** maishsk_ has joined #openstack-infra10:26
*** maishsk_ has quit IRC10:31
*** tnovacik has quit IRC10:31
*** maishsk has joined #openstack-infra10:32
*** garyk has joined #openstack-infra10:33
*** dtantsur|brb is now known as dtantsur10:33
openstackgerritJulien Danjou proposed openstack-infra/project-config: Publish Gnocchi to PyPI
*** teran has joined #openstack-infra10:35
*** mrmartin has quit IRC10:36
*** fhubik_afk has joined #openstack-infra10:37
*** teran_ has joined #openstack-infra10:38
*** teran__ has joined #openstack-infra10:40
*** fhubik has quit IRC10:41
*** teran has quit IRC10:41
*** teran has joined #openstack-infra10:42
*** teran_ has quit IRC10:44
*** kashyap has joined #openstack-infra10:44
kashyapHi, anyone remember top off their heads what's the oldest version of QEMU that's being run in the Gate?10:45
kashyapMaybe I should check in one of the CI run logs.10:45
*** teran__ has quit IRC10:45
openstackgerritSalvatore Orlando proposed openstack-infra/elastic-recheck: Add query for bug 1448067
openstackbug 1448067 in neutron "Failed to shutdown neutron" [Undecided,New]
*** sdake_ has quit IRC10:53
mordreddtroyer, lifeless: it's not that bad actually - what we'd do is publish a new thing to pypi called openstackclient - and then publish a new release of python-openstackclient that is an empty project except has a depend on openstackclient10:56
mordreddtroyer: we did it for the quantum/neutron rename, and that had _way_ more intricate depends10:58
*** dguerri is now known as _dguerri10:59
*** isviridov_away is now known as isviridov11:00
*** _dguerri is now known as dguerri11:00
*** maishsk has quit IRC11:00
*** _nadya_ has quit IRC11:00
*** dhritishikhar has quit IRC11:03
*** hashar is now known as hasharLunch11:04
*** markvoelker has joined #openstack-infra11:07
AJaegerjd__: ping11:07
*** zhiwei has quit IRC11:08
sdaguesalv-orlando: so isn't actually the bug any more11:09
openstackLaunchpad bug 1285323 in grenade "Services fail to shut down on the old side of Grenade" [Undecided,In progress] - Assigned to Sean Dague (sdague)11:09
*** hdd has quit IRC11:10
*** markvoelker has quit IRC11:12
*** e0ne is now known as e0ne_11:13
*** hdd has joined #openstack-infra11:15
*** yamamoto has quit IRC11:15
openstackgerritSean Dague proposed openstack-infra/elastic-recheck: Add a query for bug 1446583
openstackbug 1446583 in oslo-incubator "services no longer reliably stop in stable/kilo" [Critical,Fix committed] - Assigned to Julien Danjou (jdanjou)11:16
*** ddieterly has joined #openstack-infra11:16
*** alexpilotti has joined #openstack-infra11:18
openstackgerritGiulio Fidente proposed openstack-infra/tripleo-ci: Bump up delorean to allow for openstack-dashboard installation
*** cdent has quit IRC11:20
*** ddieterly has quit IRC11:20
*** dguerri is now known as _dguerri11:20
AJaegersdague: Good morning! I had to rebase the doc requirements patch, could you +2 again, please?
*** wznoinsk has quit IRC11:22
openstackgerritMerged openstack-infra/elastic-recheck: Add a query for bug 1446583
openstackbug 1446583 in oslo-incubator "services no longer reliably stop in stable/kilo" [Critical,Fix committed] - Assigned to Julien Danjou (jdanjou)11:22
*** BharatK has quit IRC11:23
AJaegerthanks, sdague11:23
*** e0ne_ has quit IRC11:23
*** dguitarbite has quit IRC11:24
*** dguitarbite has joined #openstack-infra11:25
*** _nadya_ has joined #openstack-infra11:27
*** marcusvrn has quit IRC11:28
*** marcusvrn has joined #openstack-infra11:28
*** yamahata has quit IRC11:31
*** _dguerri is now known as dguerri11:37
openstackgerritMerged openstack-dev/hacking: Updated from global requirements
*** ldnunes has joined #openstack-infra11:40
openstackgerritMerged openstack-infra/os-loganalyze: Add support for non-text files
openstackgerritMerged openstack-infra/os-loganalyze: Tidy up generators into contained objects
*** e0ne has joined #openstack-infra11:43
*** dprince has joined #openstack-infra11:49
*** fhubik_afk has quit IRC11:50
openstackgerritSean Dague proposed openstack/requirements: create a separate section for pinned requirements
*** funzo has joined #openstack-infra11:51
*** dguitarbite has quit IRC11:53
salv-orlandosdague: do we have a new one? I used that because I did a search on lp and it kind of matched the failure mode fairly well11:54
*** markvoelker has joined #openstack-infra11:54
salv-orlandook it's 1446583 then11:55
*** tnovacik has joined #openstack-infra11:55
sdaguethe logging changed a little bit with the grenade refactor so you needed "die" in there as well11:57
openstackLaunchpad bug 1448067 in neutron "Failed to shutdown neutron" [Undecided,New]11:57
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: puppet: use a single image for CI
salv-orlandoI ignored bug 1446583 so I thought there were different manifestations11:57
openstackbug 1446583 in oslo-incubator "services no longer reliably stop in stable/kilo" [Critical,Fix committed] - Assigned to Julien Danjou (jdanjou)11:57
salv-orlandoI guess this second bug can be marked as duplicate then11:57
*** otter768 has joined #openstack-infra11:58
sdagueyeh, so that was actually kind of a different bug11:58
sdaguethat was just a code bug in grenade that slipped through11:58
jd__AJaeger: pong11:59
sdaguewhen we started upgrading from stable/kilo instead of stable/juno11:59
sdaguethat was fixed with -
salv-orlandosdague: thank I will close this second bug then11:59
sdaguesalv-orlando: just did12:00
sdaguewith the review posted12:00
*** jerryz has quit IRC12:00
salv-orlandosdague: cool. I should remember that pressing f5 on your browser window is a good thing to do!12:00
*** hdd has quit IRC12:00
sdagueyeh, sorry, I've been knee deep in this bug for a week, and was just trying to figure out ways past it. I did realize this was blocking neutron yesterday, left a message in channel early on to hopefully alert people to it.12:01
*** EmilienM has quit IRC12:01
*** EmilienM has joined #openstack-infra12:01
sdagueso if you've got neutron patches that failed on grenade in the last 2 days, I'd say just blanket recheck them now12:02
*** otter768 has quit IRC12:02
*** scheuran has joined #openstack-infra12:06
openstackgerritJulien Danjou proposed openstack-infra/project-config: Publish Gnocchi to PyPI
*** kgiusti has joined #openstack-infra12:09
openstackgerritJon-Paul Sullivan (jp_at_hp) proposed openstack-infra/zuul: Reorder zuul-cloner checkout strategy
openstackgerritThierry Carrez proposed openstack-infra/release-tools: Update rccut and rcdelivery to match Kilo process
*** dboik_ has quit IRC12:10
AJaegerjd__: I commented in gerrit already, let me review your newest patch now...12:10
jd__AJaeger: yeah thanks for your help12:10
*** dizquierdo has quit IRC12:11
*** mpaolino has joined #openstack-infra12:11
*** hdd has joined #openstack-infra12:11
openstackgerritSerg Melikyan proposed openstack/requirements: Add python-muranoclient to requirements
openstackgerritMerged openstack-infra/release-tools: Use fill method to simplify string wrapping
openstackgerritSerg Melikyan proposed openstack/requirements: Add python-muranoclient to requirements
openstackgerritSerg Melikyan proposed openstack/requirements: Add python-muranoclient to requirements
openstackgerritMerged openstack-infra/release-tools: Fix lp project to repo name conversion
*** maishsk has joined #openstack-infra12:15
*** ddieterly has joined #openstack-infra12:16
openstackgerritMerged openstack-infra/release-tools: warn instead of failing with missing bug url
AJaegerfungi, sdague, pleia2: Could you review and approve today, please? The translation import worked fine on openstack-manuals but not on the other repos...12:17
sdagueAJaeger: +212:18
AJaegerthanks, sdague !12:20
*** ddieterly has quit IRC12:21
AJaegerfungi, good morning to you and thanks as well!12:21
*** nfedotov has joined #openstack-infra12:21
*** tnovacik has quit IRC12:22
openstackgerritMerged openstack-infra/irc-meetings: Add Neutron QoS Meeting
openstackgerritMerged openstack-infra/irc-meetings: Add Monasca Meeting details
openstackgerritMerged openstack-infra/irc-meetings: Add OpenStack Community Infrastructure Meeting
openstackgerritMerged openstack/requirements: Remove doc repositories from projects.txt
openstackgerritMerged openstack/requirements: Recap ceilometerclient for stable/juno
openstackgerritMerged openstack-infra/irc-meetings: Add translation test meeting
openstackgerritMerged openstack-infra/irc-meetings: Add Install guide meeting
*** hdd has quit IRC12:25
openstackgerritMerged openstack-infra/irc-meetings: Add Networking Gudie update hangout
openstackgerritMerged openstack-infra/irc-meetings: Add the Win the Enterprise meetings
openstackgerritMerged openstack-infra/project-config: Publish Gnocchi to PyPI
*** _nadya_ has quit IRC12:26
*** dboik has joined #openstack-infra12:27
*** dboik has quit IRC12:27
*** baoli has joined #openstack-infra12:27
*** dboik has joined #openstack-infra12:27
openstackgerritMerged openstack-infra/irc-meetings: Add Akanda Team Meeting
openstackgerritMerged openstack-infra/irc-meetings: Add Neutron Service Chaining Meeting
openstackgerritMerged openstack-infra/irc-meetings: Add log WG meeting
*** devvesa has quit IRC12:29
cloudnullquick question, the console logs from seem to be throwing an ISE. is there a maintenance going on? Example:
cloudnullit seems to only effect "console.html"12:29
*** Hal has joined #openstack-infra12:30
*** Hal is now known as Guest4785712:30
*** zhiwei has joined #openstack-infra12:30
openstackgerritMerged openstack-infra/project-config: Fix translation compress for manuals
fungisdague: looks like we just merged a couple of changes to os-loganalyze, could one of those be causing that do you think?12:35
*** dboik_ has joined #openstack-infra12:35
openstackgerritAndreas Jaeger proposed openstack-infra/project-config: Decrease PO files for imports (horizon+friends)
fungii'll start digging deeper in a bit if nobody beats me to it12:35
*** baoli has quit IRC12:35
*** dboik_ has quit IRC12:35
cloudnullthanks fungi !12:36
*** isviridov is now known as isviridov_away12:36
*** dboik_ has joined #openstack-infra12:36
*** dguitarbite has joined #openstack-infra12:36
*** isviridov_away is now known as isviridov12:37
*** dboik has quit IRC12:38
sdaguefungi: url?12:38
sdaguefungi: it could be12:38
sdaguecan we spot revert and see?12:39
sdagueor can you look at the logs?12:39
sdagueyeh, it looks like all console.html files are dead12:39
sdagueyeh, all kinds of files are dead12:40
*** baoli has joined #openstack-infra12:40
sdaguefungi: I think having apache server logs here is kind of required12:40
*** yolanda has quit IRC12:40
openstackgerritSean Dague proposed openstack-infra/os-loganalyze: Revert "Tidy up generators into contained objects"
ttxyep, just stumbled on same problem, 50012:41
sdagueso, until fungi gets back with apache errors, it's a guessing game12:41
ttxI guess... Moon phase12:42
*** bknudson has quit IRC12:42
sdagueno, I approved some things this morning that I assumed were more tested than they actually were12:42
ttxsdague: and I just found the motivation to look into gate-barbican-devstack-dsvm and see why it fails 60% of the time12:42
ttxI guess I'll recehck it in the mean time :)12:43
*** isviridov is now known as isviridov_away12:46
BobBallkrtaylor / asselin: Ping :) Could you let me know when you're around?12:48
sdagueok, well I'm going to start reverting the patches one at a time and see12:48
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Have get_image_name return an image_name
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
openstackgerritMonty Taylor proposed openstack-infra/shade: Remove REST links from inventory metadata
openstackgerritSean Dague proposed openstack-infra/os-loganalyze: Revert "Add support for non-text files"
openstackgerritMerged openstack-infra/os-loganalyze: Revert "Tidy up generators into contained objects"
*** dkranz has joined #openstack-infra12:49
openstackgerritSean Dague proposed openstack-infra/os-loganalyze: Revert "Add support for non-text files"
*** isviridov_away is now known as isviridov12:50
*** xyang1 has joined #openstack-infra12:50
*** mwagner_lap has quit IRC12:51
*** isviridov is now known as isviridov_away12:51
AJaegerfungi, does not yet exist since the new job merged. I noticed that we have Elastic Research already published but not mentioned, so wrote . Nothing urgent for review - we still miss some projects.12:52
*** _nadya_ has joined #openstack-infra12:52
*** e0ne is now known as e0ne_12:53
*** rlandy has joined #openstack-infra12:53
openstackgerritMerged openstack/requirements: Bump eventlet version to 0.17.3
*** e0ne_ is now known as e0ne12:55
*** afazekas has quit IRC12:55
*** dboik_ has quit IRC12:55
*** dboik has joined #openstack-infra12:56
openstackgerritOpenStack Proposal Bot proposed openstack-dev/hacking: Updated from global requirements
*** dboik has quit IRC12:56
*** bswartz has quit IRC12:56
fungiAJaeger: right the job that creates and uploads the index only fires when the files which create it are modified (file pattern restrictions for it in layout.yaml) so i'll probably need to manually trigger that unless we have any pending changes for tox.ini or other matching files12:57
*** mrmartin has joined #openstack-infra12:57
*** dboik has joined #openstack-infra12:57
AJaegerfungi: exactly - so 177077 will be such a trigger12:57
fungiAJaeger: also the individual documents only get uploaded when their respective projects have commits merge to them, so some of them may just not have a high rate of code churn12:57
*** _nadya_ has quit IRC12:57
marcusvrnhi all! Does someone know what is the openstack Jenkins CI version?12:57
sdaguefungi: did you find anything in the apache logs, or should I just keep pushing reverts until we get back to something good?12:57
fungisdague: looking at the server logs now12:58
*** alaski is now known as lascii12:58
AJaegerfungi: Yeah, we need to check which ones are missing and somehow push them...12:58
fungii was stepping away from the computer just as cloudnull spotted the error12:58
cloudnullyea, sorry about that :)12:59
sdaguenothing wrong with spotting errors12:59
sdagueI reverted the top of 2 patches12:59
fungiAttributeError: 'module' object has no attribute 'get_file_generator'12:59
sdaguefigured I'd wait until top of the hour to figure out if it deployed the fix13:00
fungii'll clean up a full traceback13:00
sdagueok, so the patch I reverted and landed removed that13:01
sdaguewhen will the revert take?13:01
*** funzo has quit IRC13:01
*** jamesmcarthur has joined #openstack-infra13:02
fungiwith clarkb's speedups, puppet lag is down to ~half an hour now13:02
openstackgerritMonty Taylor proposed openstack-infra/shade: Add thread sync points to Task
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a docstring to the Task class
fungibut i can pull it earlier manually13:03
mordredharlowja_away, Shrews, SpamapS: ^^ my original comments on harlowja_away's patch were wrong and he was right. we do need that code13:03
sdaguefungi: yeh, if you could make sure it's refreshed, that would be nice13:03
*** yamamoto has joined #openstack-infra13:04
*** gordc has joined #openstack-infra13:04
*** ajo has quit IRC13:05
fungisdague: that seems to have fixed it--thanks!13:06
*** radez_g0n3 is now known as radez13:07
*** dizquierdo has joined #openstack-infra13:08
*** yamamoto has quit IRC13:08
AJaegersdague: thanks!13:10
*** doug-fish has joined #openstack-infra13:13
*** afazekas has joined #openstack-infra13:13
*** weshay has quit IRC13:13
*** dustins has joined #openstack-infra13:13
*** ajo has joined #openstack-infra13:14
*** eharney has joined #openstack-infra13:15
fungii think civs just got 'stacked13:16
fungii'm having timeouts reaching it13:17
fungithere it goes, finally13:17
*** maishsk has quit IRC13:21
*** mpaolino has quit IRC13:22
AJaegerfungi: I just voted fine...13:22
AJaegeris everybody voting directly? ;)13:23
*** zz_jgrimm is now known as jgrimm13:23
*** mpaolino has joined #openstack-infra13:23
*** esker has quit IRC13:23
fungii suspect a thundering herd among the people who do it immediately just to get it out of the way13:23
*** peristeri has joined #openstack-infra13:25
*** jgrimm is now known as zz_jgrimm13:25
AJaegeryeah, so that they don't forget and vote one hour after it closes...13:25
*** dhritishikhar has joined #openstack-infra13:25
*** bswartz has joined #openstack-infra13:25
openstackgerritThierry Carrez proposed openstack-infra/release-tools: Update rccut and rcdelivery to match Kilo process
*** xyang1 has quit IRC13:27
*** mattfarina has joined #openstack-infra13:27
*** xyang1 has joined #openstack-infra13:28
sgordoni have a dumb Q, sorry if the wrong place to ask it, a while back we created with the help of infra a stackforge/telcowg-usecases project in gerrit (yay!) - when i view all projects in gertty however it doesn't show up13:28
*** kaisers1 has quit IRC13:28
sgordonim wondering if this is a symptom of us missing something during project setup13:28
AJaegersgordon: it exists and has content:
*** kaisers has quit IRC13:28
AJaegerSo, just a gertty problem?13:29
AJaegersgordon: jeblair has written gertty, you might want to wait another hour to ask him13:29
openstackgerritDavid Shrewsbury proposed openstack-infra/shade: Update flavor API for new get/list/search API
*** kaisers has joined #openstack-infra13:30
*** kaisers1 has joined #openstack-infra13:30
sgordonAJaeger, maybe but i dont have this issue with any other new repos13:31
sgordonthey show up automagically13:31
*** yolanda has joined #openstack-infra13:31
fungisgordon: new repos may not show up without a gertty restart? i haven't tested that13:33
*** erlon has joined #openstack-infra13:33
*** AJaeger has quit IRC13:34
openstackgerritDavid Shrewsbury proposed openstack-infra/shade: Update flavor API for new get/list/search API
*** ayoung_ has joined #openstack-infra13:34
*** mriedem_away is now known as mriedem13:35
*** ddieterly has joined #openstack-infra13:38
sdagueclarkb: when you get up -,cm13:39
openstackgerritMerged openstack/requirements: Relax pysendfile requirement
*** tiswanso has joined #openstack-infra13:43
*** bknudson has joined #openstack-infra13:44
openstackgerrityolanda.robla proposed openstack-infra/system-config: Move manage_projects logic to puppet-jeepyb
*** ayoung_ is now known as ayoung13:46
openstackgerritSean Dague proposed openstack/requirements: create a separate section for pinned requirements
openstackgerritSean Dague proposed openstack/requirements: relax final pins
*** smccully has quit IRC13:48
*** jamesmcarthur has quit IRC13:49
*** smccully has joined #openstack-infra13:49
openstackgerritEmilien Macchi proposed openstack-infra/project-config: zuul/layout: make gate-puppet-.*-puppet-syntax-future voting
*** armax has joined #openstack-infra13:50
openstackgerritMerged openstack-infra/devstack-gate: add liberty to feature grid
*** apmelton_away is now known as apmelton13:50
*** nikhil_k-afk is now known as nikhil_k13:51
*** johnthetubaguy is now known as zz_johnthetubagu13:52
*** armax has quit IRC13:52
*** kashyap has left #openstack-infra13:53
*** armax has joined #openstack-infra13:53
*** mwagner_lap has joined #openstack-infra13:53
openstackgerritZhenguo Niu proposed openstack/requirements: Update python-cinderclient to 1.2.0
*** mriedem has quit IRC13:56
sdaguefungi: is there any way to make apache logs available for situations like the os-loganalyze fail in some protected way? just because I was debugging blind on that.13:56
*** esker has joined #openstack-infra13:57
*** stevemar has joined #openstack-infra13:57
odyssey4mehi guys - a note on hpcloud vs rax images - it seems that hpcloud has no backports apt source included, whereas rax does13:57
*** dansmith is now known as superdan13:59
*** otter768 has joined #openstack-infra13:59
*** emagana has quit IRC14:00
*** mpaolino has quit IRC14:00
*** deepakcs has quit IRC14:00
*** dguerri is now known as _dguerri14:01
*** yolanda has quit IRC14:01
*** tqtran has joined #openstack-infra14:01
*** asselin_ has joined #openstack-infra14:01
*** mriedem has joined #openstack-infra14:01
*** sigmavirus24_awa is now known as sigmavirus2414:03
*** alex_xu has quit IRC14:03
*** heyongli has quit IRC14:03
*** otter768 has quit IRC14:04
sdagueodyssey4me: that's ... possible14:04
sdagueodyssey4me: is there a specific issue?14:04
*** signed8bit has joined #openstack-infra14:04
*** _dguerri is now known as dguerri14:05
odyssey4mesdague we're noticing it for the swift erasure encoding apt packages which come from the backports repository14:05
*** jp_at_hp has joined #openstack-infra14:05
mordredthis is in unittests or devstack/functional/integration?14:06
*** ChuckC_ has joined #openstack-infra14:06
mordredbecause we're VERY close to having devstack nodes all be the same across all clouds14:06
openstackLaunchpad bug 1448152 in openstack-ansible trunk "swift erasure encoding library installation failures" [Undecided,New]14:06
mordredit'll be slightly longer for unittest nodes to be the same14:06
*** jp_at_hp has quit IRC14:06
*** jp_at_hp has joined #openstack-infra14:06
*** jp_at_hp has joined #openstack-infra14:06
*** ChuckC has quit IRC14:06
jp_at_hpCan I get some feedback on from anyone please? mordred - would you know who would be best to talk to about zuul-cloner?14:07
mordredjp_at_hp: clarkb should be up soon14:08
*** sks has quit IRC14:08
*** hasharLunch is now known as hashar14:08
*** afazekas has quit IRC14:09
*** jp_at_hp1 has joined #openstack-infra14:09
mordredjp_at_hp: although looking at that change, you'll probably want feedback from sdague and jeblair too14:09
jp_at_hp1thanks mordred14:11
clarkbI'm reading the old code and the order the commit msg has and dont see the difference14:11
*** vhoward has left #openstack-infra14:11
*** jp_at_hp has quit IRC14:11
*** vhoward has joined #openstack-infra14:11
clarkbwell nevermind then?14:12
clarkboh two clients14:12
clarkbits too early14:12
clarkbzuul ref -> specified branch -> master is what the old code does14:12
clarkbhow is thia different?14:12
*** amitgandhinz has joined #openstack-infra14:13
mordredthe old code at least claims to do "zuul ref for specified branch -> zuul ref for master -> tip of specified branch -> tip of master"14:13
*** ChuckC_ has quit IRC14:13
*** armax has quit IRC14:13
mordredthe new node claims to do "zuul ref for specified branch -> tip of specified branch -> zuul ref for master -> tip of master"14:13
*** armax has joined #openstack-infra14:14
clarkbzuul ref for master isnt a thing iirc. only zuul ref14:14
mordredwell, the comments in the old code list it as a thing14:14
clarkbyou get a unique ref branch not multiples14:14
* mordred has not read the actual flow - is just reading the comments14:14
*** alex_xu has joined #openstack-infra14:14
*** heyongli has joined #openstack-infra14:14
mordredfungi: ManagerStoppedException: Manager hpcloud-b1 is no longer running14:14
mordredfungi: do you happen to know what might cause that?14:14
mordredfungi: hp is investigating times where we fail to boot nodes or time them out and delete them - and found a couple of places where the node booted properly and then we deleted it14:15
fungimordred: um... no? but i'll see if i can identify a list of potential causes14:15
clarkboh wait nvm thats how grenade works so I am probably wrong14:15
mordredfungi: in investigating the given node, 72fdbe6b-f240-4237-93df-952ba99151a4, I found that to be the cause for deleting it14:16
mordredclarkb: moar kofffeee14:16
sdaguemordred: what's the use case here, I'm concerned it will do the wrong thing with upstream14:16
*** scheuran has quit IRC14:16
mordredsdague: meet jp_at_hp1 . jp_at_hp1 meet sdague14:16
* mordred is just a matchmaker14:16
clarkbah I see it does it for botg zuul refs in the old code too14:17
*** tonytan4ever has joined #openstack-infra14:17
sdaguebecause zuul ref's should be uniqueish, right?14:17
clarkbso I am not really sure what the desired change of behavior is14:17
sdaguemtreinish: if that's fiddler on the roof, I'm going to have to hit you14:17
clarkbjp_at_hp1 can you expand on that?14:17
mtreinishsdague: heh, then I'll have to hide14:17
clarkbgoing by comment diff its flipping 2 and 314:18
fungiso apparently a nodepool.task_manager.TaskManager object starts out running when created, and looks like that only changes when its stop() method is called14:18
*** Ala has quit IRC14:18
fungiso looking for places calling stop on it14:18
jp_at_hp1clarkb sdague - so...  When performing CI on a product that uses multiple repositories to create it's final built artefacts, we want to run CI in the gate in a way that is most consistent.14:19
openstackgerritMerged openstack-infra/shade: Have get_image_name return an image_name
*** zz_johnthetubagu is now known as johnthetubaguy14:20
fungiinside nodepool.NodePool.reconfigureManagers there's a call to stop a manager when reconfiguring14:20
jp_at_hp1If I try and take a simple example of 2 repos A and B, each of which build on different branches, master and not-master14:20
*** Ala has joined #openstack-infra14:20
sdaguejp_at_hp1: so that assumes that *every* repo that's being used with zuul has the same branch structure14:20
fungimordred: is it possible the incidents you found happened during a reconfiguration?14:20
jp_at_hp1sdague: not at all14:20
*** esker has quit IRC14:21
sdaguejp_at_hp1: so what happens if you are testing projects1, 2, 3. refs A (on 1), B on (2), C on (3)14:21
sdagueon branch foo14:21
fungisdague: on making apache logs public, not sure. are there good log scrubbers out there that can sanitize those properly?14:21
sdaguebut C doesn't have branch foo14:22
jp_at_hp1so if I make a change to repo A and a change to repo B, on the master and not-master branches respectively, what I want to happen in the gate is that my test of the later change runs with both those zuul refs.  In both the old and new code that will happen.14:22
sdagueyou then don't test C, right?14:22
sdaguehonestly, I think we need jeblair for this. I can never keep the whole model in my head.14:23
jp_at_hp1sdague: typing too slow.  sometimes human conversation is easier :(14:23
*** wenlock has joined #openstack-infra14:23
openstackgerritMerged openstack-infra/shade: Remove REST links from inventory metadata
jp_at_hp1sdague: I don't think your example is quite true.14:23
*** mikap has joined #openstack-infra14:24
jp_at_hp1I think I can show that through my example...14:24
*** mikap has quit IRC14:24
jp_at_hp1so in your example, ref C on (3) is on some branch other than foo?14:25
*** hdd has joined #openstack-infra14:25
jp_at_hp1or is it specifically on the fallback branch of 'master'?14:25
*** woodster_ has joined #openstack-infra14:26
*** tqtran has quit IRC14:26
clarkbI don't think it matters, it will checkout the indicated branch in either case14:26
*** mtanino has joined #openstack-infra14:27
clarkbthe scenario that is important to consider is what ref of B is checked out when testing C14:27
clarkb(I think)14:27
*** ChuckC_ has joined #openstack-infra14:28
jp_at_hp1in my example of 2 repos14:28
jp_at_hp1A and B, each with master and non-master as the branch we want to use14:28
jp_at_hp1when I build I want to always use those 2 branches.14:29
jp_at_hp1A on master and B on non-master14:29
clarkbthat breaks an assumption that zuul makes14:29
sdagueright, so you assume that all projects have the same branch structure14:29
clarkb(the always bit)14:29
jp_at_hp1I don't see that as true at all14:30
clarkbyes you can do overrides to change the indicated branch14:30
jp_at_hp1OK - I know what the problem is14:30
openstackgerritDoug Hellmann proposed openstack-infra/release-tools: Update to set defaultbranch
fungimordred: looks like there was a new provider manager created for each of the hpcloud providers at 2015-04-23 17:48 which is also when 72fdbe6b-f240-4237-93df-952ba99151a4 failed to be added14:30
clarkbbut zuuls always default behavior is to test a change to B on non master against A on non master14:30
*** dguitarbite has quit IRC14:30
*** yolanda has joined #openstack-infra14:30
jp_at_hp1When talking about the zuul-cloner, I am explicitly thinking of using the mapping ability to choose branches for things.14:30
jp_at_hp1As I read the docs, zuul doesn't care about branches14:31
clarkbzuul absolutely cares about branches14:31
jp_at_hp1it makes a unique ID for every change in the pipeline, which is all projects and all branches in the pipeline14:31
clarkbjp_at_hp1: but then it only hands out a single zuul_branch to the job14:32
jp_at_hp1then the cloner makes the choice about what to test through the use of the mapping table first, zuul_branch (where the main change under test is) second14:32
clarkbso what it sounds like you would like to do is override zuul_branch in all cases to change the indicated branch14:32
*** esker has joined #openstack-infra14:32
jp_at_hp1no - this is getting quite off topic.  And in fact even here I can show incorrect behaviour of current code, I think...14:32
clarkbjp_at_hp1: I am trying to understand the use case so that I understand your example14:33
*** zz_jgrimm is now known as jgrimm14:33
jp_at_hp1stable branches14:33
clarkbis ^ not something that you are doing? force A to master regardless and B to non master regardless?14:33
jp_at_hp1that is a use-case that is broken by current code14:33
clarkbstable branches are working for us just fine I think14:33
jp_at_hp1the current code will do the wrong thing14:33
clarkbbut it may be there is a corner case when you do the above overriding14:33
openstackgerritDoug Hellmann proposed openstack-infra/release-tools: Add script to update .gitreview files
clarkb(also we aren't using this with a large set of stable branches yet, its mostly master to master everywhere iirc)14:34
jp_at_hp1no - I'm pretty sure that (based off the comment) the current code *should* be broken.,14:34
clarkbso back to an example14:34
clarkbA change to master is pushed, B depends on A change to stable/foo is pushed14:34
jp_at_hp1yes - I'm 100% sure that in majority of cases my change is corerct.  Really I want to know what the use case of the current code is14:34
sdagueso... we're running 3 different stable branches upstream, + 3 different combinations of 2 branch jobs in grenade. And we're testing the right things.14:35
clarkbzuul-cloner will checkout B at stable/foo/zuulref, zuul-cloner will checkout A at stable/foo/zuulref14:35
clarkbjp_at_hp1: ^ that should work14:35
clarkbnow if A does not have a stable/foo it will try to checkout master/zuulref14:35
clarkbthat should also work14:36
sdagueso it would be good to know what is working around this at our level if you think the code is going to give the wrong results14:36
clarkbsdague: well we aren't zuul clonering grenade yet but the intent is to do so14:36
sdagueclarkb: oh, gotcha14:36
openstackgerritMatt Riedemann proposed openstack-infra/elastic-recheck: Generalize fingerprint for horizon bug 1447476
openstackbug 1447476 in OpenStack Dashboard (Horizon) "No module named pathlib" [Critical,Fix released]
clarkbif A has been forced to use master via an override we should still checkout B at stable/foo/zuulref and A at master/zuulref14:37
jp_at_hp1clarkb: you're not quite right there  (unless it was just the comment that threw me off14:37
clarkbjp_at_hp1: I feel like I am missing the fourth use case to make your change clock14:37
clarkber click14:37
clarkbjp_at_hp1: indicated branch will be master, zuul will checkout master/zuulref before anything else for that project14:38
jp_at_hp1repo A we want not-master14:38
clarkbsure we can s/master/foobranch/ and I htink the answer is still the same14:38
jp_at_hp1So if there is a change going through on A/master, but not on A/not-master, when we test we will be testing using rpo A/master/zuulref where what we *actually* want is A/not-master14:39
jp_at_hp1precedence in first order is branch, second order if there is a zuulref for that branch14:40
clarkbjp_at_hp1: so thats actually a question for zuuls mergers and not the cloner14:40
jp_at_hp1no - that is the code in the cloner that I fixed14:40
clarkbjp_at_hp1: if the merger makes a ref for A/not-master then your example is wrong14:40
clarkband I don't know if it does that or not14:40
jp_at_hp1no - it doesn't make a ref for A/not-master, because there is no change going through the gate on that repo14:41
clarkbiirc it will only if there is a change in the list on that branch somewhere14:41
openstackgerritMatt Riedemann proposed openstack-infra/elastic-recheck: Generalize fingerprint for horizon bug 1447476
jp_at_hp1but if a change is going through on A/master, it will make a ref for that, right?14:41
openstackbug 1447476 in OpenStack Dashboard (Horizon) "No module named pathlib" [Critical,Fix released]
clarkbjp_at_hp1: yes it definitely makes a ref on any branch for which there is a change14:41
jp_at_hp1and then the job using the cloner will pull the master ref, which it doesn't want, instead of the indicated branch.14:41
clarkb is what I am looking at now to see if this behavior is preexisting14:43
clarkb appears to be the key bit there14:44
clarkbwhich is mimicked on lines ~139 of the old side of your change's diff14:44
jp_at_hp1clarkb: cool - that function does exactly what my patch does...14:44
jp_at_hp1so that basically says my patch is correct?14:45
clarkbjp_at_hp1: not sure yet14:45
clarkbjp_at_hp1: I think zuul-cloner attempts this behavior14:45
jp_at_hp1well, I read that code, and my change makes zuul-cloner behave exactly like that function14:45
clarkbbecause on 139 it checks if the branch exists14:45
clarkbso in your example it should say yes I have branch stable/foo then make that the fallback zuul ref branch, it will fail to fetch that because no ref is made14:46
jp_at_hp1look at the final conditional block in the set_project function14:47
clarkbthen it should hit the else on 161 and checkout the fallback branch even though there was a master zuul ref14:47
clarkbso I think zuul-cloner already does what you want it to there14:47
clarkbjp_at_hp1: but you never hit that else14:48
clarkbon 14214:48
clarkbso master isn't even considered14:48
jp_at_hp1So either the comment or reading of the code is wrong14:48
clarkbas long as you have properly indicated you want a different branch and that branch exists the behavior you want appears present14:48
clarkbthe comment is correct14:49
*** zz_dimtruck is now known as dimtruck14:49
clarkbits just not going to be evaluated completely on each run14:49
jp_at_hp1no - the comment is wrong14:49
phschwartzHas anyone seen issues with jjb when not using the absolute latest jenkins stable where runs fail with a 400 when doing a job create?14:49
clarkbso it might be better to write a slightly more verbose comment with better branching14:49
*** ashleighfarnham has joined #openstack-infra14:49
jp_at_hp1it doesn't consider a zuul ref on master ahead of the tip of the indicated branch14:49
*** jamesmcarthur has joined #openstack-infra14:49
jp_at_hp1and I'm still not 100 % convinved :(14:50
clarkbjp_at_hp1: it does if the indicated branch does not exist14:50
jp_at_hp1gimme 5 to read the code again.14:50
clarkbjp_at_hp1: so there is a branch14:50
clarkbwhich is why I think there is confusion, its not a fall through set of steps14:50
jp_at_hp1clarkb: the comment is this:14:50
jp_at_hp1The reference lookup is attempted in this order:        1) Zuul reference for the indicated branch  2) Zuul reference for the master branch   3) The tip of the indicated branch        4) The tip of the master branch14:51
jp_at_hp1that is plain wrong if the actual behaviour is correct14:51
clarkbyes and 2) is IFF indicated branch does not exist14:51
clarkbotherwise its still correct14:51
jp_at_hp1And I'll be happier to just correct the comment, if it is behaving properly ;-)14:51
clarkber 4) is also IFF the indiicated branch does not exist14:51
openstackgerritMerged openstack-infra/elastic-recheck: Generalize fingerprint for horizon bug 1447476
openstackbug 1447476 in OpenStack Dashboard (Horizon) "No module named pathlib" [Critical,Fix released]
*** vhoward has quit IRC14:51
jp_at_hp1yes.  So the comment *is* *wrong*.  Phew - I'm not mad.14:52
clarkbits incomplete14:52
clarkbthe order is correct14:52
openstackgerritMerged openstack-infra/elastic-recheck: Add query for bug 1448067
*** vhoward has joined #openstack-infra14:52
openstackbug 1448067 in grenade "Failed to shutdown neutron" [Undecided,Fix released] - Assigned to Sean Dague (sdague)14:52
clarkbbut there is a second condition that determines whether or not things are evaluated at all14:52
*** arxcruz has quit IRC14:53
jp_at_hp1clarkb: shades of wrong are still wrong14:53
clarkbsureI am not saying don't fix it14:53
clarkbbut the thesis we started with is the order was wrong14:53
*** samuelBartel has joined #openstack-infra14:53
clarkbwhih is not the case14:53
*** julim has joined #openstack-infra14:54
jp_at_hp1true - but that has nothing to do with the fact the comment is wrong ;-)14:54
*** sabeen has joined #openstack-infra14:55
jp_at_hp1Glad the code is correct.  I think a few comments to make it more decipherable will be good - I'll amend my review.14:55
*** nelsnelson has joined #openstack-infra14:57
*** nelsnelson has quit IRC14:58
*** _nadya_ has joined #openstack-infra14:59
*** dangers_away is now known as dangers15:00
*** doug-fish has left #openstack-infra15:01
jp_at_hp1clarkb: maybe I should just change my commit message to say it makes the existing code more readable? ;-)15:01
openstackgerritMatt Riedemann proposed openstack-infra/elastic-recheck: Remove query for bug 1348204
openstackbug 1327218 in OpenStack Compute (nova) "duplicate for #1348204 Volume detach failure because of invalid bdm.connection_info" [High,Fix released] - Assigned to Matt Riedemann (mriedem)15:01
clarkbjp_at_hp1: yes, especially since the order change there isn't quite right either15:02
clarkbcomment should read something like if indicated_branch exists: 1) try zuul ref for indicated branch 2) checkout indicated branch else 1) try zuul ref for master 2) checkout master15:03
*** achanda has joined #openstack-infra15:03
jp_at_hp1yeah - I have changed that comment, that's easy15:03
jp_at_hp1but the actual code just seems so over-complicated for no particular reason15:04
clarkbya it may be simpler to basically write it as ^15:04
jp_at_hp1trying to add comments to it to describe what it is actually doing in a sane way is making my head hurt15:04
*** Bsony_ has joined #openstack-infra15:04
*** Bsony has quit IRC15:06
jp_at_hp1so...  fallback_zuul_ref is totally unnecessary if you check for the branch before getting it15:06
jp_at_hp1that also simplifies the if statement at the end15:06
jp_at_hp1because the flow is...15:06
clarkbya you  can just make zuul_ref and set it appropriately based on the existence of the branch15:07
jp_at_hp1check for upstream branch, if not set branch to master15:07
jp_at_hp1check for zuul ref on branch15:07
jp_at_hp1use branch15:07
*** davideagnello has quit IRC15:07
jp_at_hp1clarkb: ^^^ that is the change to make, isn't it?15:07
*** achanda has quit IRC15:08
clarkbI guess we are assuming upstream branch is the predetermined indicated branch15:08
clarkbso ya15:08
jp_at_hp1well - the code setting indicated branch is good, and won't change15:08
clarkbright its the checking for its existence after and setting the branch to operate on and the zuul ref15:09
jp_at_hp1ok - new patch in 5 - simples15:09
*** sandywalsh has quit IRC15:10
*** dkranz has quit IRC15:11
*** dhritishikhar_ has joined #openstack-infra15:12
*** e0ne is now known as e0ne_15:12
*** dhritishikhar has quit IRC15:12
*** dkranz has joined #openstack-infra15:12
*** jamesmcarthur has quit IRC15:12
openstackgerritMatthew Treinish proposed openstack/requirements: Add cap for tempest-lib so reqs syncs don't break the world
*** baoli has quit IRC15:13
*** e0ne_ is now known as e0ne15:13
*** sabeen3 has joined #openstack-infra15:14
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: Abort failed cherry-pick and reverts
*** doug-fish has joined #openstack-infra15:15
*** sabeen has quit IRC15:15
*** dannywilson has joined #openstack-infra15:16
*** Bsony_ has quit IRC15:17
openstackgerritMerged openstack-infra/elastic-recheck: Remove query for bug 1348204
openstackbug 1327218 in OpenStack Compute (nova) "duplicate for #1348204 Volume detach failure because of invalid bdm.connection_info" [High,Fix released] - Assigned to Matt Riedemann (mriedem)15:18
*** dhritishikhar_ has quit IRC15:18
*** david-ly_ is now known as david-lyle15:22
*** ihrachyshka has quit IRC15:23
*** hashar has quit IRC15:23
openstackgerritJulien Danjou proposed openstack-infra/project-config: Run Gnocchi devstack tests for stable/1.0 against Kilo
*** nfedotov has quit IRC15:27
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a method for getting an endpoint
openstackgerritMonty Taylor proposed openstack-infra/shade: Make keystone endpoints work like the others
openstackgerritMonty Taylor proposed openstack-infra/shade: Add early service fail and active check method
*** cody-somerville has quit IRC15:29
*** e0ne is now known as e0ne_15:29
fungiheading out to lunch... back in a bit15:30
fungibtw, last call for infra outreachy internship ideas and mentor volunteers
*** e0ne_ is now known as e0ne15:30
*** rm_work|away is now known as rm_work15:30
fungiwhen i get back from lunch i'm going to try to get up with vkmc about next steps15:30
fungipleia2: jeblair: clarkb: mordred: SergeyLukjanov: jhesketh: everyone else ^15:31
mordredfungi: I'm having lunch with jeblair in about 30 minutes - he's otherwise offline today15:31
*** jlvillal has quit IRC15:31
fungimordred: i expected so15:31
mordredfungi: I'll try to grab suggestions from him at lunch?15:32
fungiwe've been warned ;)15:32
fungiyes, please!15:32
fungiokay, back in a while15:33
*** marcusvrn has quit IRC15:34
openstackgerritSpencer Krum proposed openstack-infra/system-config: Convert hiera configuration to support public data
*** liam_ has joined #openstack-infra15:39
*** liam_ is now known as Guest6792615:39
*** Swami has joined #openstack-infra15:39
anteayamorning zaro15:40
*** cody-somerville has joined #openstack-infra15:40
*** andreykurilin__ has joined #openstack-infra15:40
*** vlaza has quit IRC15:41
*** marcusvrn has joined #openstack-infra15:41
*** yolanda has quit IRC15:43
*** Guest67926 has quit IRC15:43
*** Bsony has joined #openstack-infra15:43
*** ilyashakhat has quit IRC15:43
nibalizerclarkb: you might find that interesting ^^ implements a really old spec15:44
*** ilyashakhat has joined #openstack-infra15:44
*** MarkAtwood has joined #openstack-infra15:44
*** cdent has joined #openstack-infra15:45
*** MaxV has quit IRC15:45
*** _nadya_ has quit IRC15:46
*** cody-somerville has quit IRC15:48
mordrednibalizer: so - that assumes that there will be one and only one hieradata dir yeah?15:48
anteayafungi: I have added some thoughts15:48
mordredand it will be in /opt/system-config15:48
nibalizermordred: yes15:48
mordrednibalizer: does hiera do symlinks well?15:49
nibalizerif that doesn't jive with whatever else is floating in /opt/system-config let me know15:49
*** marcusvrn has quit IRC15:49
nibalizermordred: i think it can follow them ya15:49
zaroclarkb: just read the gerrit post, that's good news.15:49
nibalizercould test pretty easy15:49
mordrednibalizer: like, can we make /opt/system-config/hieradata a symlink to /etc/puppet/hieradata15:49
clarkbzaro: well fungi pointed out why it may be bad news too :) but it is good that mina may not be at fault15:49
clarkbzaro: that may be an exploitable DoS15:49
mordrednibalizer: main question there being that I'd be slightly scared to have a git checkout managed by puppet have a directory inside of it that wasn't part of that git repo and that losing would be bad15:50
mordrednibalizer: but if we can symlink, then ossum15:50
nibalizermordred: huh so is /opt/system-config the repo or is that just a container and /opt/system-config/production is the repo?15:50
mordredoh - one sec15:50
mordredyah. I suck15:51
mordrednibalizer: you're right - /opt/system-config/production is the repo15:51
*** sdake has joined #openstack-infra15:51
*** marcusvrn has joined #openstack-infra15:51
mordrednibalizer: so this would put a common.yaml, a group and an fqdn dir in the root of the system-config repo, yeah?15:52
nibalizeri think hiera/{those things}15:52
nibalizerat least thats what i meant to do15:52
mordredoh - wow. I cannot read15:52
mordredyup. I see it15:53
*** dtantsur is now known as dtantsur|afk15:53
nibalizerah yea i screwed up the comments at the bottom15:53
openstackgerritSpencer Krum proposed openstack-infra/system-config: Convert hiera configuration to support public data
mordredyour public data comment at the bottom15:53
nibalizerthe fix ^15:53
mordredyup. that reads well15:53
mordredI like it15:53
mordredit might be worthwhile adding a comment to the commit message that applying this change will need a manual maint task to be performed15:54
mordredactually - we could add a symlink right now, and then at some point if we like it, we could move the hieradata dir to /opt if we wanted15:54
mordredso that the patch can land whenever15:54
mordrednibalizer: actually - can we put environment in the private one too? we have those15:55
thingeeclarkb: this landed. I think we spoke about the 0.1.0 tag having to manually published to pypi?
mordredso "hierdata/%{::environment}/common"15:56
mordrednibalizer: ^^15:56
nibalizermordred: totes15:56
* nibalizer will respin15:56
mordrednibalizer: that way I can just do ln -s /etc/puppet/hieradata /opt/system-config15:56
clarkbthingee: yup we will need to trigger those jobs for you15:56
clarkbpleia2: when you wake is doing ^ something you are interested in?15:56
clarkbpleia2: I can walk you through it15:57
openstackgerritSpencer Krum proposed openstack-infra/system-config: Convert hiera configuration to support public data
thingeeclarkb, pleia2 thanks :)15:58
mordrednibalizer: +2 from me - nicely done!15:59
mordrednibalizer: I knew that one of these days we'd find the elegant way to do it15:59
nibalizerty, ty its been a while since that spec has been looked at15:59
*** mriedem is now known as mriedem_nom15:59
*** marcusvrn has quit IRC16:00
*** otter768 has joined #openstack-infra16:00
zaropelix: what is your opinion on this?
clarkbso thats a dangerous change16:00
*** asselin_ has quit IRC16:00
*** andreykurilin__ has quit IRC16:00
clarkbbecause we use default hiera values for testing, then we move the hiera data away16:01
*** jlanoux has quit IRC16:01
mordredclarkb: what's dangerous about that/16:01
*** baoli has joined #openstack-infra16:01
clarkbnibalizer: mordred what is going to move the hiera data? just a manual step?16:01
mordredclarkb: yah16:01
clarkbmordred: we will reset all our DB passwords to known values16:01
clarkbmordred: because hiera will say "no data here" then use the default16:01
mordredoh - so - I've already made a symlink16:02
mordredso there will be zero seconds of time when the data is not in a location that we know about16:02
mordredbecause doing things in a different order would be suck16:02
clarkbok, I am just calling it out because that change does not call it out16:03
clarkband our downstream users may want to know this16:03
nibalizerclarkb: i dont understand your concern yet16:03
clarkbsince site.pp is apparently a thing others use16:03
mordredcool - just making sure we were worried about the same thing16:03
mordrednibalizer: if the change lands without the corresponding hieradata move16:03
clarkbnibalizer: if we change the source of hiera data without also having data in that location the default hiera lookup values will be used16:03
mordrednibalizer: then a puppet run could happen in production that would wind up setting things to defaults16:03
nibalizerah yea16:03
clarkbnibalizer: this will reset all database passwords and do a swarm of other nastyness16:03
nibalizerwell cp works too16:04
nibalizerfollowed by rm16:04
*** baoli_ has joined #openstack-infra16:04
anteayaI'm doing some archeology work in the interest of untangling nova-net and neutron, some conversations took place on the mailing list in the first half of 2012 and our archives only go back to May 2012:
clarkbsure, point is how to make sure that we don't pwn a bunch of downstreams (and ourselves though mordred seems to have taken care of us)16:04
anteayamight there be any other location where the -dev list was archived?16:04
clarkbanteaya: on launchpad16:04
samueldmqfungi, ping - you around ?16:04
clarkbanteaya: I think it was called openstack-dev there as well16:04
*** otter768 has quit IRC16:04
samueldmqfungi, iirc, you said me days ago that we had experimental jobs for keystone v3, am I right ?16:05
zaropelix: it's a big jjb change, my concern is that unit test might not be enough.16:05
*** marcusvrn has joined #openstack-infra16:05
nibalizerclarkb: i think this is a systemic problem with passing those defaults in via site.pp ... ill do some thinking on a way to not do that16:05
*** baoli has quit IRC16:06
odyssey4meit would appear that the zuul status page isn't updating16:06
* mordred does not really like the passing of defaults, just for the record16:06
*** Ala has quit IRC16:07
anteayaclarkb: thank you16:07
clarkbodyssey4me: it appears to be updating for me (note the two queue counters in the top left, zuul is chewing on a backlog of events)16:07
*** e0ne has quit IRC16:07
odyssey4methe jobs do appear to be transitioning between jenkins and gerrit, so perhaps this is a graphite issue?16:07
anteayaclarkb: yes:
clarkbmordred: nibalizer ya maybe we can just stop doing that entirely16:07
odyssey4meah ok clarkb cool - just wanted to give a heads-up16:07
*** smccully has quit IRC16:08
openstackgerritSean Dague proposed openstack/requirements: update openstackclient requirement to 1.0.3
pelixzaro: sure, guess the question is what additional tests do we want to add to give the level of confidence?16:09
openstackgerritMerged openstack-infra/project-config: Make oslo.db py3 job voting
*** gyee has joined #openstack-infra16:09
anteayaI think I should go for a walk before heading into the dust of the past16:09
pelixzaro: I think the performance benefits are definitely worth investing in adding threading code to JJB, it's just a question of what is needed to ensure correct behaviour16:10
openstackgerritMerged openstack-infra/project-config: Using Neutron network by default in gate-murano-congress-devstack-dsvm
*** ZZelle has quit IRC16:11
clarkbnibalizer: mordred two comments16:11
*** smccully has joined #openstack-infra16:12
*** marun has joined #openstack-infra16:12
*** sandywalsh has joined #openstack-infra16:12
*** dprince has quit IRC16:13
clarkbtime for breakfast, pleia2 ping if/when you are interested in triggering zuul post jobs directly16:13
openstackgerritJon-Paul Sullivan (jp_at_hp) proposed openstack-infra/zuul: Make the cloner strategy clearer
*** jogo is now known as flashgordon16:18
*** tjones1 has joined #openstack-infra16:19
asselinBobBall, hi16:20
jp_at_hp1clarkb: thanks a mill for the discussion earlier - I have an updated patch up that I think makes things clearer...16:20
*** unicell has quit IRC16:20
*** tjones2 has joined #openstack-infra16:21
*** tjones1 has quit IRC16:22
*** rm_work is now known as rm_work|away16:22
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a method for getting an endpoint
openstackgerritMonty Taylor proposed openstack-infra/shade: Make keystone endpoints work like the others
openstackgerritMonty Taylor proposed openstack-infra/shade: Add early service fail and active check method
openstackgerritMonty Taylor proposed openstack-infra/shade: Replace e.message with str(e)
openstackgerritMonty Taylor proposed openstack-infra/shade: Call super in OpenStackCloudException
BobBall#asselin: XenProject has been accepted for a BoF for the summit around the setting up and use a of 3rd party CI.  The original submission has been lost (not quite sure how) so we wanted to take the opportunity to re-imagine the BoF, given the recent developments as well; we were thinking it would be most useful to have a group of us talking about and sharing what we each think are the best practices for setting one up.  What are your t16:24
*** dizquierdo has quit IRC16:28
zaropelix: i've tested and it looks good.  do you feel the unit tests are adequate?16:29
asselinBobBall, what's BoF?16:29
*** tjones2 has quit IRC16:29
asselinand your comment got cut off. "What are your t"16:29
asselinBobBall, sorry back in a bit16:30
BobBallBirds of a feather; My understanding it's just a group of people with the same interests getting together to talk about a particular topic (i.e. not a formal presentation etc)16:30
*** tjones1 has joined #openstack-infra16:30
*** tjones1 has quit IRC16:30
BobBallasselin: Just said "What are your thoughts on that" :)16:30
*** tjones1 has joined #openstack-infra16:30
*** jcoufal has quit IRC16:31
*** tjones1 has joined #openstack-infra16:32
*** tsg_ has joined #openstack-infra16:32
*** tjones1 has left #openstack-infra16:32
*** patrickeast has joined #openstack-infra16:32
*** shashankhegde has joined #openstack-infra16:32
*** Guest47857 has quit IRC16:32
*** jcoufal has joined #openstack-infra16:33
*** jcoufal has quit IRC16:36
*** jistr has quit IRC16:36
krtaylorBobBall, we have been talking about a third party CI working group session for weeks, here is the etherpad ->
rakhmerovhi fungi, I wasn't able to attend the last infra team meeting but I read its log. So what do I need to do to move mistral to openstack space?16:36
krtaylorBobBall, it is proposed as a cross-project session ->
pelixzaro: I think it's good enough, unless we can think of something else to test right now, I'd vote for adding the patch (once the conflict is resolved) and asking people to stress it a bit to shake out edge cases that we can't forsee right now.16:37
BobBallYes; I've seen that and having a formal session is great.  I was wondering if having a BoF on just sharing what each of us has done and what we think should be the best practices would be a useful complement to the working group session.  My understanding of the original BoF submission was that it is effectively superceeded by that session16:37
krtaylorBobBall, part of this is being discussed as a infra design session (downstream-puppet) with asselin driving it16:37
krtaylorBobBall, a BoF is fine, but that is exactly the focus of the working group session16:39
krtaylorBobBall, if the cross-project session is not selected, then a BoF would be an excellent backup16:39
krtaylorBobBall, I do not envision the cross-project session to be a formal "presentation" format, a working session or worst case fish bowl if too big for all at one table16:40
*** yamamoto has joined #openstack-infra16:42
BobBallI saw a BoF to be maybe covering one of the 'improved ways to share' what we've each been doing but of course, as discussions often do, I can see that many of the things in that proposed session would be of interest.16:42
*** unicell has joined #openstack-infra16:42
*** wenlock_ has joined #openstack-infra16:42
krtaylorI do not know what ttx schedule is for choosing cross-project sessions, but I'd prefer that to get decided first16:43
krtaylorBobBall, is the BoF a done deal?16:43
BobBallIt's been approved to happen, yes16:43
openstackgerritJulien Danjou proposed openstack-dev/pbr: Support Unicode in requirements file
krtaylorBobBall, ah, ok, well then, if I might suggest that a show and tell for best practices be directed toward one of the working group meetings16:44
*** dimtruck is now known as zz_dimtruck16:45
*** nelsnelson has joined #openstack-infra16:45
krtaylorBobBall, we have a "Highlighting a System" section of the meeting that is geared up to share these topics16:45
krotscheckclarkb: So, after thinking about it, I actually agree with mordred that pbr should be aware of javascript things. To me it's less a question of "hey pbr should only do python things", and more a question of "package maintainers already trust pbr, and thus anything submitted there will be held to a higher standard to maintain that trust". For example, by enforcing the --single_version_externally_managed flag16:45
*** nelsnelson has quit IRC16:46
krtaylorBobBall, I'd suggest using f2f time to work out details of how to share, and brainstorming  - just a thought16:46
*** yamamoto has quit IRC16:46
krotscheckclarkb: Also, because a "Python web app" is actually a thing people build frequently16:46
clarkbI was more concerned about the reasonableness aspect. particularly how to get the js back out again16:47
clarkbwe can certainly be opinionated about that as pbr is full of opinions but I am not sure there is a good answer to that question16:47
*** baoli_ has quit IRC16:47
*** baoli has joined #openstack-infra16:48
*** derekh has quit IRC16:48
*** tjones2 has joined #openstack-infra16:48
clarkbdo we not pull it out and expect webapps to refer to it with pkg_resources and serve it directly? or do we make a default install location on disk? options seems not good16:49
clarkbputting the js in the sdist is straightforward, run bower the include as package data16:49
BobBallkrtaylor: I have to say I'm getting confused the more I think about it; it seems that the BoF is to be part of the main summit schedule not design summit, but clearly the appropriate audience for a 3rd party CI discussion is more the devs than users... *confused*16:51
morganfainbergsamueldmq, we had experimental jobs for python3 in keystone not keystone v3 afaik16:51
clarkbkrotscheck do you have plans for the install step?16:51
*** achanda has joined #openstack-infra16:52
krotscheckclarkb: I'm currently writing a hook which checks for the existence of a bower.json file and adds the necessary logic to resolve those dependencies on pip install.16:52
clarkbif the plan is to make package data then have python serve it, would it make more sense to just commit the files and sdist as package data?16:52
krotscheckclarkb: yes.16:53
krotscheckclarkb: YES.16:53
*** zz_dimtruck is now known as dimtruck16:53
krotscheckclarkb: But rock and hard place.16:53
clarkbok then why the pbr change?16:53
*** sdake has quit IRC16:53
samueldmqmorganfainberg, hmm, maybe I have confused myself then16:53
*** Swami_ has joined #openstack-infra16:53
samueldmqmorganfainberg, I thought I had talked about that with fungi16:53
*** davideagnello has joined #openstack-infra16:54
krotscheckclarkb: Because I'm exploring options for how to make javascript things packageable, and if this is the path we end up going down I want to make sure it's feasible.16:54
krotscheckclarkb: I'm doing my research for vancouver.16:54
*** bnemec is now known as beekneemech16:54
clarkbkrotscheck also keep in mind there are two steps to install, sdist to make a tarball (this is the easy bit) and the install step of copying files to where they will live on disk16:54
*** dannywilson has quit IRC16:55
*** Swami has quit IRC16:55
*** dannywilson has joined #openstack-infra16:55
krotscheckclarkb: Yeah, the various install steps aren't that well documented, so right now I've made the assumption that 'install' is the place to do that.16:55
clarkbI think the install command implies sdist and runs it for you16:56
krotscheckclarkb: Where this logic lives is certainly up for discussion. At the moment I need to have this hook make a best-effort to detect/install node & npm.16:56
krotscheckclarkb: And then it needs to invoke 'npm install', which resovles dependencies and anything else configured in the package.16:56
*** pblaho has quit IRC16:57
krotscheckclarkb: The assumption is that after that's done, all the files will be where they need to be.16:57
clarkbthats a bad assumption16:57
morganfainbergsamueldmq, once devstack can deploy a v3-only keystone we will do an experimental job16:57
clarkbthey will end up in the sdist16:57
morganfainbergsamueldmq, then we can chase down all the other fixes needed.16:57
rakhmerovsorry guys for jumping in.. clarkb: do you know who can help with stackforge -> openstack renaming?16:57
krotscheckclarkb: *waves hands around*16:57
clarkbbut apache/nginx/etc wont know how to get at those files to serve them16:58
*** sarob has joined #openstack-infra16:58
krotscheckclarkb: Ok, if you were writing this hook, and the end goal is "pip install ." results in a fully functional web app, what would you do?16:58
*** nelsnelson has joined #openstack-infra16:58
krotscheckclarkb: (With the caveat that whoever runs pip install needs to have reasonable clue on how to setup a vhost/etc.16:58
clarkbkrotscheck put files in an sdist so that the sdist step works. this allows ypu to pypi16:58
clarkbbut I have no good answer for installing the js at install time16:59
samueldmqmorganfainberg, well, to deploy a v3 only keystone is just to remove the /v2.0 apps in paste-ini16:59
samueldmqmorganfainberg, that's how I did manually16:59
morganfainbergsamueldmq, no there is a lot of devstack stuff that uses v2 apis even if v3 is specified in stack.sh16:59
clarkbthe simplest approach is to force applications to serve the js out of their packages16:59
morganfainbergsamueldmq, i'm mostly done with a fix to solve this.16:59
morganfainbergsamueldmq, but it's whack-a-mole16:59
pleia2clarkb: sorry, have a bit of a cold so I'm slower moving this morning, have you triggered the zuul jobs yet?16:59
*** gmurphy has left #openstack-infra16:59
* fungi is back. the party can resume ;)17:00
morganfainbergpleia2, no! Feel better!17:00
fungianteaya: thanks for the input!17:00
morganfainbergpleia2, not allowed to be sick! :)17:00
clarkbbut that enforces an opinion on hosting static files that I expect most to dislike17:00
clarkbpleia2 I have not17:00
pleia2morganfainberg: my husband was home from work 3 days because of it, I had hoped to retain immunity, alas!17:00
clarkbalso feel better17:00
samueldmqmorganfainberg, great ! you're coming with devstack and I am coming with the working services with v3 (just testing so far)17:00
morganfainbergpleia2, oh gah! seriously feel better!!17:00
fungisamueldmq: i don't recall mentioning experimental keystone jobs for either v3 api or python 3.x, but i also have a terrible memory17:00
samueldmqmorganfainberg, and we will converge soon :)17:00
pleia2clarkb: I'm feeling ok, let's walk through this :)17:00
openstackgerritJon-Paul Sullivan (jp_at_hp) proposed openstack-infra/zuul: Make the cloner strategy clearer
morganfainbergfungi, we had py3 for a while.. we removed them because $REASONS or made them expirimental17:01
samueldmqfungi, k, maybe I got confused :p sorry17:01
krotscheckclarkb: Is there a place where the order of commands is documented?17:01
morganfainbergfungi, but i expect we will be py3 compat in liberty (fully). we are ~2 libraries away afaict17:01
morganfainbergboth with py3 friendly alternatives17:01
clarkbkrotscheck: I have no idea, setuptools (and distutils) are dark magic and I find most of the details are undocumetned17:02
clarkbkrotscheck: setuptools does have docs though17:02
*** tsg_ has quit IRC17:02
clarkbkrotscheck: anyways I think an important thing is to have the sdist step package up the files so that if someone installs off of pypi they do not need npm and node and bower, they just need python17:02
*** tsg_ has joined #openstack-infra17:02
krotscheckclarkb: I agree.17:02
*** vhoward has left #openstack-infra17:03
krotscheckclarkb: And hopefully the sdist command will check for the necessary tools.17:03
fungirakhmerov: if nobody's talked to you yet, mainly you just need to git grep in the openstack-infra/project-config repo for places where your project name is mentioned, and then update any occurrences of stackforge with openstack and submit it for review. also link that review url on our meeting agenda wiki page so we can discuss it at one of our meetings when we plan our next gerrit maintenance (which17:03
fungiwon't be until kilo is released, so as not to impact the release process at this crucial stage)17:03
*** marcusvrn1 has joined #openstack-infra17:04
rakhmerovfungi, I have already done all that17:04
krotscheckKnow what? I'm going to let my subconscious chew on this one until monday and do storyboard things instead.17:04
*** marcusvrn has quit IRC17:04
clarkbkrotscheck: for the sdist step you would implement a hook into there (either as a pbr hook or in pbr directly) that makes sure all of the files are present then makes sure the files are added to the package as package data17:04
fungiclarkb: krotscheck: you might be interested in the recent thread on distutils-sig about classifying python-packaged file types (for fhs support, but more generally so that platform-specific deployment locations can be declared by the platforms where the package is being repackaged or installed)17:04
clarkbkrotscheck: that should get everything into the pckage.17:05
*** yolanda has joined #openstack-infra17:05
clarkbbut then on the install side you have to decide where those files go17:05
clarkbdstufft did say warehouse is just serving the js directly from python and not putting the files anywhere outside of the pckage itself17:05
rakhmerovfungi, the patch is and at the last meeting I was told to go to this channel directly and talk to some of you about the further steps17:05
rakhmerovso I'm here17:05
clarkbthat will work as long as your python application is aware that it is its job to serve the static files and that they live in its package data17:06
krotscheckclarkb: Well, there are different patterns there, which the setup config file should handle.17:06
*** shashankhegde has quit IRC17:06
krotscheckclarkb: Something like a config block with a install_my_static_files_here option or something17:06
*** ssam2 has quit IRC17:07
fungirakhmerov: yep, those were the further steps17:07
fungirakhmerov: now we just need to wait for an opportunity to schedule the work around it17:07
clarkbkrotscheck: I am not sure if you can feed config at install time. Or rather I don't think that has been implemented anywhere so don't have good prior art17:07
*** [HeOS] has quit IRC17:07
rakhmerovfungi, ok, I'll be waiting17:07
*** HeOS has joined #openstack-infra17:08
fungirakhmerov: please add it at
clarkbkrotscheck: pbr can probably do it but I would be surprised if setuptools had direct support for that17:08
*** dimtruck is now known as zz_dimtruck17:08
krotscheckclarkb: pbr does it via hooks.17:08
rakhmerovfungi, ok, I'll do right away17:08
fungirakhmerov: after the meeting i updated the agenda moving your topic to the renames section, but at the time didn't know the url17:08
krotscheckclarkb: I believe the config block is passed through. I might be wrong17:08
clarkbkrotscheck: pbr packages don't take any config at install time iirc17:08
rakhmerovfungi: got it17:09
clarkbkrotscheck: the config is baked at sdist17:09
fungirakhmerov: so where i have "project-config review link goes here" just replace that with the url17:09
*** mriedem_nom is now known as mriedem_rmorse17:09
*** mriedem_rmorse is now known as mriedem17:09
*** tsg_ has quit IRC17:10
*** tnovacik has joined #openstack-infra17:10
rakhmerovfungi: done17:10
*** tsg_ has joined #openstack-infra17:10
krotscheckclarkb: Well, given that distros end up rebuilding their own things anyway, I'm not too concerned about that. Their version of insert-pythonlib-here could install to /var/www, anothers to /lib/www, etc. The pip version... well, that's up to the package maintainer really.17:11
krotscheckI mean the code maintainer17:11
clarkbkrotscheck: its not distros you worry about its anyone doing `pip install refstack`17:11
pleia2clarkb: assuming it's the right script, I have notes on using /opt/zuul/tools/ but crafting the appropriate trigger job is something I'm still learning17:11
clarkbbecause that will operate on the sdist (or whl) on pypi17:11
clarkbpleia2: kk let me pull of the tag and see what the params should look like17:12
*** tjones2 has quit IRC17:12
*** sdake has joined #openstack-infra17:12
krotscheckclarkb: Yep, not worried about that. Installing something like refstack is going to require additional configuration anyway, which means file locations will need to be documented no matter what.17:12
*** rm_work|away is now known as rm_work17:12
pleia2clarkb: the easy bits, --job publish-to-pypi --project openstack/os-brick --pipeline post (right?)17:13
clarkbpleia2: close, the job is actually going to be the tarball job first, we then wait for the tarball to be on then we run the pypi upload job17:13
krotscheckclarkb: And since the preferred install of something like that is in a private venv anyway, anyone configuring the webserver has access to set the uid/gid on the files anyway and can make sure insert-webserver-of-choice can host it.17:13
clarkbpleia2: we also have to feed it the tag as the ref iirc17:13
krotscheck(Unless they're installing as root, in which case I have no pity for them)17:14
pleia2clarkb: nods17:14
clarkbkrotscheck: sure, you can install it that way and rely on documentation to have people configure webservers17:14
clarkbor node or whatever is using the js17:14
*** jp_at_hp1 has quit IRC17:15
clarkbpleia2: my only other real concern is I think there may be a different script to use beacuse the fixes for something never got commited to git :/17:15
clarkbpleia2: but I think that isn't actually a problem17:16
pleia2clarkb: does seem pretty reliant on having git hashes17:17
clarkbpleia2: ya I think we feed it the git hash of the commit the tag points to, pbr then generates the version properly because of git17:17
clarkbone sec making the command now17:18
*** dprince has joined #openstack-infra17:18
pleia2aha, publish-to-pypi is a template, need to break it down into it's jobs for this17:18
*** tjones1 has joined #openstack-infra17:19
pleia2cinerama: whenever you're around, I can approve 176578 and we'll keep an eye on how it does17:20
fungipleia2: clarkb: btw if you grep through my .bash_history on zuul for you'll see plenty of release pipeline examples17:20
openstackgerritClint 'SpamapS' Byrum proposed openstack-infra/infra-manual: Add clarification on questions in reviews
clarkb`/opt/zuul/tools/ --job os-brick-tarball --project openstack/os-brick --pipeline release --refname 0.1.0 --newrev 0b89d9b3e7d899f318eadbeb866f6f675f20ead0 then we need a logpath17:21
clarkbfungi: oh thats handy17:21
cineramapleia2: i am here17:21
*** bhuvan has joined #openstack-infra17:21
clarkbah the newrev should be for the tag itself and refname is refs/tags/0.1.017:22
clarkbpleia2: ok constructing the actual command for tarball job now17:22
*** ociuhandu has joined #openstack-infra17:22
pleia2clarkb: ok, once you have it, we can maybe walk through where you got each value so I can do a better job of looking them up myself17:23
*** zz_dimtruck is now known as dimtruck17:23
funginormally when you're retriggering a job which already ran, it's a lot easier because you can just look in jenkins for the parameters which were passed in the previous run17:24
* pleia2 nods17:24
*** dguerri is now known as _dguerri17:24
clarkbsudo /opt/zuul/tools/ --url= --job os-brick-tarball --project openstack/os-brick --pipeline release --newrev 89a6b0ae7a61ffa697555306b154119915f0cac0 --refname refs/tags/0.1.0 --logpath 89/89a6b0ae7a61ffa697555306b154119915f0cac0/release/os-brick-tarball/manual-run-117:24
clarkbyup definitely can walk through what each of those things is17:24
fungiso while i _do_ know how to figure out the various values to pass, it's often easier to just cut-n-paste them from the failed/broken/whatever run (this case where the job never ran/existed at all is a little harder)17:25
*** _dguerri is now known as dguerri17:25
pleia2clarkb: do we need a --url ?17:26
clarkbpleia2: url is the base url to fetch the ref from. it will be joined with the value of project to get the full path17:26
*** salv-orlando has quit IRC17:26
clarkbpleia2: job is the name of the job, you can work backwards from the zuul layout to find what the name is for that I actually hopped on jenkins.o.o and searched for os-brick because that was quick17:26
clarkbproject is the project name and needs the org prefix thing17:26
clarkbpipeline is release because that is the pipeline these jobs will run in the zuul layout17:27
clarkbyou have to work back from the template to get that17:27
*** Krinkle|detached is now known as Krinkle17:27
pleia2clarkb: oh, I see the url now, in my previous work with this it was a zm0* url, not review17:27
clarkbpleia2: ya we can't use zm0X because these jobs were never triggered and won't have refs on any zm0X, but gerrit iwll have them so we use it instead17:27
fungiright, normally zuul will have asked a merger to prepare the refs for a job17:28
fungijust not in this case17:28
clarkbpleia2: newrev is the sha1 ofthe tag, not the sha1 of the commit the tag points to. You can get that from git by doing a `git show-ref`17:28
fungiluckily for post/tag/pre-release/release pipelines the zuul ref is of little consequence since it should be identical to something in gerrit or our git mirrors17:28
clarkbpleia2: I figured that out by looking at fungi's examples on zuul.o.o17:28
clarkbI also figured out that the refname needed the full path of refs/tags/0.1.0 by looking at that example17:29
fungiyou can also look at the jenkins parameters from recent runs of another similar job for a different repo for more examples17:29
*** melwitt has joined #openstack-infra17:29
clarkbpleia2: finally the log path is where to copy the files, this is mostly arbitrary but best to make it follow the pattern the job normally would've taken. That is first_2_of_sha1/sha1/pipeline/jobname/someuuid17:30
clarkbthis really makes me wnat to write a patch to zuul enqueue to handle releaes17:30
pleia2ok, confirming now that I can look all of these up properly and get the same results as you17:30
*** cdent has quit IRC17:30
clarkbpleia2: since the tarball job must run before the upload job we can start with the above command then when it is done update the command to run the upload job17:31
clarkband we can confirm that the tarball is on before running the upload as well17:31
* pleia2 nods17:32
clarkbsince enqueue doesn't grok all this release stuff yet an short term thing to do may be to push a patch to the trigger script with better help text for each of these fields17:33
pleia2clarkb: ok, easy enough, all lgtm17:34
pleia2I can run this on zuul.o.o now if we're all good17:34
clarkbpleia2: yup go for it17:34
pleia2running, no errors yet, but I think --url=http... should have been url http...17:38
clarkbpleia2: I think argparse will do the correct thing for us either way17:39
pleia2finished ok, and we now have
clarkblogs don't show up, so lets find the jenkins console log before the next thing17:39
clarkb(its likely because we didn't pass in swift stuff17:39
pleia2ok, just start searching through the jenkinsXXs?17:40
clarkbya :/ is the base link and just increment the number as you go17:40
clarkb06 is the winner looks like17:40
fungipassing in swift stuff would i think in this case also mean doing the credential issuing dance zuul normally performs, so likely not worth trying17:40
clarkband confirmed that swift log uploads bailed out because we didn't pass the info (second to last line of the console log)17:41
*** tqtran has joined #openstack-infra17:41
clarkbfungi: ++ for these sorts of cleanups I am not super worried about it17:41
fungiand yeah, i can't think of any uses of the trigger-job script which wouldn't better be served by adding support to the enqueue subcommand17:41
fungioh, other than "testing gearman"17:42
*** shashankhegde has joined #openstack-infra17:42
clarkbpleia2: I have also confirmed the tarball and whl are present at
*** r-daneel has joined #openstack-infra17:42
clarkbso I think we are ready to run the next job which is the one that wil upload to pypi17:42
clarkbfor that we want to run the 'os-brick-pypi-both-upload' job17:43
pleia2similar command but --job os-brink-pypi-wheel-upload ?17:43
clarkbpleia2: you should be able to just replace os-brick-tarball with ^ in the previous command17:43
cineramapleia2: feel free to pull the trigger on that17:43
clarkbpleia2: and in the log url (though that likely won't matter)17:43
pleia2ah yes, log url too17:43
cineramapleia2: worst case we can nuke everything & reinstall17:43
*** prad_ has joined #openstack-infra17:44
pleia2clarkb: both-upload or wheel-upload? (template I'm looking at has wheel)17:44
*** prad has quit IRC17:44
*** prad_ is now known as prad17:44
clarkbpleia2: for pre-release we do only wheel, for release we do both-upload17:44
openstackgerritJulia Kreger proposed openstack-infra/shade: Add update_machine method
openstackgerritJulia Kreger proposed openstack-infra/shade: Add Ironic machine power state pass-through
openstackgerritJulia Kreger proposed openstack-infra/shade: Add Ironic maintenance state pass-through
pleia2ok, so we want both-upload here17:44
clarkbpleia2: the reason for that was old pip doesn't know how to install wheels and new pip won't install pre releases by default17:45
clarkbpleia2: so by uploading wheels only for pre releases we ensure no one uses a pre release without explicitly wanting it17:45
fungihowever it turns out that plan has a downside17:45
fungipypi still shows the wheel-only prerelease if we hide all but the most recent release17:45
fungiwe've also all but abandoned use of pre-release tags anyway17:46
pleia2alright, doing this one now17:47
*** cody-somerville has joined #openstack-infra17:47
*** cody-somerville has quit IRC17:47
*** cody-somerville has joined #openstack-infra17:47
*** rm_work is now known as rm_work|away17:47
*** dhritishikhar has joined #openstack-infra17:47
*** bcourt has quit IRC17:48
pleia2seems good
clarkbyup lgtm17:49
openstackgerritJulia Kreger proposed openstack-infra/shade: Add Ironic maintenance state pass-through
clarkbpleia2: thansk for the help17:49
openstackgerritJulia Kreger proposed openstack-infra/shade: Add Ironic machine power state pass-through
*** bcourt has joined #openstack-infra17:49
pleia2clarkb: thank you!17:50
clarkbshould probably let thingee know and have cinder/brick do any checking they want to do on their side17:50
clarkbbut this looks gone to me17:50
thingeepleia2, clarkb thanks!!17:50
*** harlowja_away is now known as harlowja17:51
*** tnovacik has quit IRC17:51
pleia2cinerama: approved, fingers crossed :)17:52
harlowjamordred u were wrong and i was right on ? cool17:53
harlowjathat whole shade task thing is similar to (and similar to taskflow task also, ha)17:54
harlowjaperhaps u guys should use taskflow :-P17:54
openstackgerritMatt Riedemann proposed openstack-infra/system-config: Index screen-barbican logs
openstackgerritMike Perez proposed openstack/requirements: Adding os-brick for managing local volumes
openstackgerritMerged openstack-infra/puppet-zanata: Update to 3.6.2
clarkbharlowja: except I think this is all synchronous17:57
harlowja*for now* ha17:57
clarkbharlowja: the idea being that your api rate limit should control the execution rate17:57
openstackgerritMerged openstack-infra/project-config: Set the proper homepage for akanda repositories.
openstackgerritMerged openstack-infra/project-config: Set the proper homepage for akanda repositories.
clarkbno its intentional17:57
clarkbyou can't just async full speed ahead at oepnstack, the api explodes and badness happens17:57
*** btran has joined #openstack-infra17:58
harlowjaya, thats why is useful for folks (hopefully)17:58
harlowja (also synchronous, but using the same futures interface)17:59
harlowjaanyways, maybe ^ not relevant for u guys, either way18:00
*** dhritishikhar has quit IRC18:00
clarkbspeedutils doesn't seem to natively grok thread safe queues?18:01
*** otter768 has joined #openstack-infra18:01
harlowjaunsure what u thinking18:01
clarkbQueue.Queue.get(timeout) raises exceptions18:01
harlowjait just limits the rate at which some wrapped iterator  returns values18:01
clarkbwhich would bubble out and break your iteration18:01
harlowjaok ?18:02
clarkbI guess you can just handle that by restarting iteration after whatever timeout handling needs to happen18:02
clarkbharlowja: right it might be nice to say stopiteration cleanly in that case18:02
harlowjahmmm, think i know what u mean18:02
harlowjafuture improvements then :-P18:02
openstackgerritTristan Cacqueray proposed openstack-infra/system-config: Use common jenkins master implementation
flashgordonpleia2: for getting out the vote18:03
flashgordonpleia2: can you re-email people 24 after the polls open if they didn't vote?18:04
flashgordonas in is that possible18:04
*** dhritishikhar has joined #openstack-infra18:04
openstackgerritTristan Cacqueray proposed openstack-infra/puppet-openstackci: Create a jenkins server like
*** ZZelle has joined #openstack-infra18:05
*** otter768 has quit IRC18:05
pleia2flashgordon: there isn't currently a mechanism for that in civs, it would all be manual18:06
pleia2we can resend individual ballots one by one18:06
*** sandywalsh has quit IRC18:06
pleia2we can send reminders to the whole dev list, and I'm on twitter patrol with it (busy morning!)18:07
*** sandywalsh has joined #openstack-infra18:08
*** hemna has joined #openstack-infra18:08
*** dguerri is now known as _dguerri18:09
*** erikmwilson is now known as Guest885318:11
*** erikmwilson has joined #openstack-infra18:11
*** erikmwilson_ has joined #openstack-infra18:11
*** Swami has joined #openstack-infra18:12
*** erikmwilson is now known as Guest738318:12
*** erikmwil_ has joined #openstack-infra18:12
*** Guest7383 has quit IRC18:12
*** erikmwil_ is now known as erikmwilson18:12
*** HeOS has quit IRC18:13
*** Swami_ has quit IRC18:13
*** vinkman has joined #openstack-infra18:13
openstackgerritJoshua Harlow proposed openstack-infra/release-tools: Include 'source_url' in notes and use readme parser function
*** vinkman has left #openstack-infra18:14
*** johnthetubaguy is now known as zz_johnthetubagu18:15
*** ivar-lazzaro has joined #openstack-infra18:16
SpamapSfungi: well played (re +2 comment ;)18:16
openstackgerritMike Perez proposed openstack/requirements: Adding os-brick for managing initiators
fungiSpamapS: well played (re improving our documentation)18:17
*** ivar-lazzaro has quit IRC18:19
*** tjones1 has quit IRC18:19
*** ivar-lazzaro has joined #openstack-infra18:20
tristanCasselin: greeting sir, I'd like your opinion on the jenkins master refactor I proposed here:
openstackgerritThai Tran proposed openstack-infra/project-config: Create a repository for stackforge/xstatic-nvd3
asselintristanC, hi, thanks for the ping. I will take a look!18:21
openstackgerritThai Tran proposed openstack-infra/project-config: Create a repository for stackforge/xstatic-angular-nvd3-directives
tristanCcool, feedback are most welcome, puppet stuff are pretty new to me :)18:22
clarkbSpamapS: fungi I was hoping not to get sucked into that thread but I mostly disagree with the idea that I can't -1 because I have questions18:22
clarkbI wrote feedback on the docs update18:23
cineramapleia2: the footer still says we're on often does puppet run?18:23
pleia2cinerama: you can keep an eye on it by going to
clarkbas a core reviewer for a project it is my job to understand the changes going into it. If I don't understand them that is worthy of a -1 because something needs to be addressed. Either the proposal itself, my understanding of the code, documentation etc18:23
SpamapSclarkb: I'm glad to have your dissent, and I think through the two positions we'll find goal posts that the readers can aim to land somewhere between. :)18:23
pleia2cinerama: hm, seems we're not running puppet there regularly for some reason18:24
clarkbleft a suggestion that -1's come with concrete actionable feedback though so others can be confident if concerns have been addressed even if the original source of the -1 is no longer active/present/revieweing18:24
fungiclarkb: that sounds like concrete, actionable feedback18:24
cineramaclarkb, SpamapS: so my contention with -1 for questions is in a situation where the reviewer is asking a question about how a third-party tool or library works. it is frustrating to get held up on that18:25
SpamapSI wonder if we should also time-out -1's18:25
*** tsg_ has quit IRC18:25
clarkbfungi: I could do "I don't understand why this code change is made. Can you explain why function foo() needs to be called here?"18:25
SpamapSlike, if a -1 has been sitting there for 45 days, and author has responded, we could just remove it.18:25
fungiclarkb: and i agree, as i said in the thread, we should avoid questions for the sake of rhetoric and hyperbole, because they don't translate well in an international/cross-cultural context18:25
clarkbfungi: so not a nebulous "I don't understand" but instead a concrete "this is what I don't get, please explain it to the rest of us"18:25
pleia2clarkb, fungi - did I make a mistake/miss a step when setting up translate-dev, or did we turn off puppet somewhere along the way?18:26
cineramaclarkb, SpamapS: if the reviewer is -1 on grounds of the code in the patch itself i think that is fair enough18:26
clarkbpleia2: oh! I think we still haven't accepted the ssh key18:26
clarkbpleia2: so ansible is failing to log in18:26
clarkbpleia2: on the puppet master you can check /var/log/puppet_run_all.log iirc to check and see what ansible says about that host18:26
pleia2clarkb: thanks, checking18:26
clarkbpleia2: if that is the case you can just accept the key, eithe run ssh or use keyscan18:26
fungiclarkb: and i do think that we should probably ask that the change itself be made more clear or include appropriate commentary, rather than us being confused and having it explained only in review comments which almost nobody is likely to see again once it merges18:26
SpamapSclarkb: That sounds like a request for code documentation, and could be worded as such. :)18:27
*** salv-orlando has joined #openstack-infra18:27
clarkbSpamapS: but it may not be documentation that is necessary18:27
SpamapSclarkb: I do think we could improve my prose to make it clear that questions with specific needs are appropriate for -1.18:27
fungipleia2: yeah, our launch readme probably needs to mention that you should sudo ssh to the fqdn at the end of the launch process and accept the host key18:28
clarkbSpamapS: maybe its a matter of pointing me at learning how ansible modules work18:28
*** tjones1 has joined #openstack-infra18:28
pleia2fungi: nods, all this fancy new ansible stuff18:28
fungipleia2: as usual, our documentation lags progress18:28
clarkbSpamapS: or maybe the change itself is invalid but without prompting with a question and giving any incentive to address it nothing happens, code merges and now everything is unhappy18:29
SpamapSclarkb: Yeah, I think you've found the line and are tap-dancing on it. IMO if the answer to your question is that you need to learn about X, you kind of wasted the contributor's time and should have +0'd. If however the answer is "Oh dear I forgot to mention that this depends on Ansible modules which is implicit here, I'll add it to the commit message, and here's where the docs are for you Clark", then it was a good -1.18:29
*** tjones1 has quit IRC18:30
*** tjones1 has joined #openstack-infra18:30
*** salv-orlando has quit IRC18:30
clarkbSpamapS: and your suggestion is that we should all avoid any confrontation because -1s are being see nthat way and just use 018:31
cineramapleia2: oh by the way, we should be able to configure the log location now for wildfly, so let me know if you want it to live somewhere else and we can make that change18:31
*** tjones1 has left #openstack-infra18:31
*** patrickeast has quit IRC18:31
pleia2cinerama: I think /var/log/wildfly would make the most sense (easy on our brains if we're looking for it)18:31
*** patrickeast has joined #openstack-infra18:31
clarkbanyways, I see code review as a conversation, there is no hard rule for when to -1 or -2 or 0 or +1 or +218:31
*** tjones1 has joined #openstack-infra18:31
SpamapSclarkb: My suggestion is that we inform reviewers of the expectations, and that they use their brain to decide +0 or -1. We have not informed them of where the line is yet, so their brains have been making poor choices at times.18:32
SpamapSAnd if there's two lines, and a big grey area, that is _totally fine_18:32
SpamapSJust right now the whole thing is grey.18:32
pleia2host key accepted, next time puppet tries to run it should be ok18:33
*** dhritishikhar has quit IRC18:34
*** dhritishikhar has joined #openstack-infra18:34
*** rm_work|away is now known as rm_work18:34
clarkbcrinkle: can you expand on what you want to wait for before 176343 merges?18:34
clarkbcrinkle: comment on the change is fine18:34
crinkleclarkb: sure18:35
clarkbSpamapS: I think that by having hard fast rules we replace grey areas with well defined areas that don't fit into their alloted space on the graph18:35
*** ajo has quit IRC18:36
clarkbSpamapS: we lready observe that many different projects have different review conventions18:36
clarkbthese have grown somewhat organically and lkely address different domain specific issues18:36
SpamapSclarkb: I think I'm more willing to trade some review quality for review efficiency than you are. :)18:37
clarkbfor example infra won't approve changes to production without someone willing to babysit for example18:37
*** ayoung is now known as ayoung-bye18:37
clarkbtripleo is much more fond of -1ing for nitpicks than infra18:37
*** ayoung-bye has quit IRC18:37
greghaynesheh, not sure about that :p18:37
*** hashar has joined #openstack-infra18:37
clarkbbut these different conventions work in the different domains18:37
clarkbgreghaynes: maybe its changed18:37
clarkbgreghaynes: but statistically I think tripleo had one of the highest negative review rates of any project18:38
clarkbbut thats fine if it results in the reviews that you want to get as a project18:38
greghaynesIMO we can probably just focus on the egregiously bad case, identify that, and let people figure out how far that case extends18:38
greghayneswhich is -1 "how does this work"18:39
mordredoh my18:39
greghaynesor maybe "-1 I didnt read this explain it to me"18:39
mordredI get back from lunch and we're discussing -1 habits18:39
* greghaynes -1s mordred's lunch decision18:40
fungii have to agree, i have gotten plenty of poor -1 reviews which just say "why did you use [x] here?" without even continuing "...instead of [y]?" or "...given that it causes [z]?"18:40
*** cody-somerville has quit IRC18:40
*** dhritishikhar has quit IRC18:40
fungithis is not a programming proficiency quiz, it's code review. if you have suggestions on how i could have done it better then please provide them ;)18:40
cineramayes, i think we should try to avoid assumptions like "everyone knows [foo] is considered harmful, i shouldn't have to explain to you why"18:41
SpamapSclarkb: the nit picking in tripleo has subsided. It was a cultural thing and not actually useful.18:41
mordredclarkb, SpamapS: one of the things I'd love is if it was possible to ping someone in a review thread18:41
*** dhritishikhar has joined #openstack-infra18:41
SpamapSmordred: yeah, that would go well with the timeout I was suggesting.18:42
mordredlike, clarkb -1's something, I respond and reference him and it shows up on his queueu again18:42
SpamapSmordred: ping starts clock.. clock runs out.. -1 is removed18:42
*** jlvillal has joined #openstack-infra18:42
mordredbecause my problem is that if I leave a -1 with a question, I'll literally never see the response18:42
mordredwhich is why  Italk to most people about questions in irc18:42
pleia2cinerama: Zanata 3.6.2 \o/18:42
greghaynesI have the same issue18:42
cineramapleia2: yay we didn't break it!18:42
mordredof course, then I have to go back and add in what we learned in IRC as a comment so other people know18:43
cineramapleia2: i did test it but my install has less datur18:43
*** signed8bit is now known as signed8bit_ZZZzz18:43
pleia2cinerama: still, good work with testing18:43
SpamapSmordred: I find that to be rather natural really. "Hey we talkd elsewhere and it was awesome and we're good. +1"18:43
mordredyah. that works well18:43
mordredbut what doesn't work well at least for me is having a back and forth on a review beacuse I miss thigns18:44
mordredbut that might just be me18:44
greghaynesTheres the cases where people dont ping in IRC though, and theres really not a great way for me to realize someone replied to my question18:44
*** hodos has joined #openstack-infra18:44
mordredthat's my problem18:44
greghaynesso it gets ignored, then I feel like a jerk18:44
flashgordonfungi: epic quote18:44
mordredSpamapS: what do you think the repr in that Exception should look like?18:44
mordredflashgordon: did I miss a good fungi?18:45
flashgordonvery good18:45
SpamapSmordred: a common convention is that it should equal code that could be eval'd to reproduce a similar/the same object.18:45
*** Longgeek has quit IRC18:45
cineramapleia2: so i'm working on the proposal slave scripts & obviously i'm kind of putting somethign out there which i expect, nay, DEMAND people rip to bits so we get it right, but i'm wondering if we have showstoppers at the moment in terms of what the CLI gives us to work with18:46
SpamapSmordred: so before the super change, it did.. it was   OpenStackCloudException("noooo","99")18:46
*** _nadya_ has joined #openstack-infra18:46
cineramapleia2: for example, would love to have your eyes on it but i don't see a way to specify percentage completion of what we pull down18:47
harlowjaSpamapS clarkb i was also talking with folks internally (at y!) about having a nit-pick button (that like has weight -0.1) ; so if 10 nitpicks == -118:47
harlowjasomeone should add that :)18:47
SpamapSharlowja: no god no18:47
SpamapSpeople will get to 7 and start digging deep18:47
harlowjathe ocd people then would have a button to click18:47
SpamapSharlowja: and click and click and click and click18:47
SpamapSThey can click the close-window button for all I'm concerned18:48
harlowjamaybe the button should do that then18:48
hodosfor thirdparty ci: which backend should I test if my driver got removed????18:48
harlowjapretty sure some javascript can do that, lol18:48
SpamapSAs a recovering nit-picker.. I don't want to even acknowledge the practice. :)18:48
fungithe ocd button should also increment when you click it, so that you can count along18:48
hodosjust LVM ?18:48
harlowjaSpamapS do u go to nit-picker anonymous ?18:48
mordredSpamapS: also, I don't understand your comments about needing to send all of the arguments to super ...18:49
harlowja*as a recovering nit-picker18:49
SpamapSLike I still do reviews the way I used to, but now I go back and delete half the comments because they're not actually useful to anybody but the pedant inside me.18:49
clarkbI have just approved 176618 if there is a reason to not run that test in the gate hurry up and let me know :)18:49
SpamapSmordred: super(OpenStackCloudException, self).__init__(message, extra_data)18:49
clarkb(but it should prevent an issue we actually ran into so I would prefer running that test in the gate)18:49
mordredSpamapS: but OpenStackCloudException does not support extra_data ?18:49
asselinhodos, your driver is cinder?18:50
openstackgerritMerged openstack-infra/infra-manual: How To Avoid Merging Specific Files
SpamapSmordred: it does so18:50
*** rm_work is now known as rm_work|away18:50
mordredSpamapS: it does???18:50
hodosasselin: yes, couple of nexenta drivers for cinder. got removed due to missing CI18:50
clarkbhodos: asselin: Iwould assume you would test with your driver reapplied to the code base so that you can show the driver works with CI18:51
*** _nadya_ has quit IRC18:51
mordredSpamapS: that's an arg I added to __init__18:51
fungi"shade/tree" is so awesome btw18:51
SpamapSmordred: Exception's signature is *args18:51
SpamapSfungi: we need peach too18:51
mordredSpamapS: gotcha18:51
cineramafwiw, in the +0 reviews i've done with suggestions-i-don't-think-should-necessarily-stop-merging people generally do tend to acknowledge them18:51
SpamapSok, tummy wants blue crab burger18:53
* SpamapS goes to find it18:53
fungithat sounds amazing18:53
SpamapSfungi: it looks amazing too
fungii get crabcake sandwiches around here pretty often18:53
clarkbmtreinish: I htink you are an expert on tempest regex stuff, can you look at 176662 and tell me if the regex change there is appropriate?18:54
fungilotsa crab on/around the island18:54
clarkbmtreinish: specifically does that say all tests that match both strings? which should be no tests18:54
hodosclarkb: asselin: ok...that's an assumption or guideline ?18:55
*** erikmwilson has quit IRC18:55
reedwhere should I log a RFE for the sandbox doc? storyboard?18:55
*** erikmwilson_ is now known as erikmwilson18:55
clarkbhodos: I can't speak for cinder, but if the point is to show that you are testing your stuff then you will want to test it with that driver18:55
*** dangers is now known as dangers_away18:55
clarkbreed: sandbox doc being documetation around using git with gerrit and having sandbox be the arena?18:56
mtreinishclarkb: it gets passed as args to testr directly18:56
*** rm_work|away is now known as rm_work18:56
hodosclarkb: thanks, let me ask for cinder18:56
reedclarkb, yes :)18:56
mtreinishand testr ors all the regexes it gets18:56
mtreinishso that should work18:56
clarkbmtreinish: does that mean all tests will work or is it an XOR evaluation18:56
openstackgerritMerged openstack-infra/project-config: Run project-config-gerrit job in gate check queue
fungiclarkb: presumably
reed is probably missing some details18:57
clarkbmtreinish: er, what I am trying to ask is does that OR such that both work or XOR so that only one set does. I guess not XOR is what makes sense18:57
mtreinishclarkb: that should be all tests they want to run should work. It's OR not XOR18:57
asselinhodos, it's documented in the cinder wiki FAQ18:57
clarkbmtreinish: thanks18:57
mtreinishclarkb: you can always do a testr list-tests on tempest with the regex18:57
clarkbmtreinish: I thought list tests idn't apply the regex18:58
reedsandbox.html may offer a playground, with sandbox the git repo offering a feature branch to fool around with18:58
reedor maybe it's only me that to be thick18:58
clarkbreed: well I added a feature branch to sandox18:58
clarkbwe may need to update acls for the merge commits18:59
reedbefore I forget, I'd like to add a bug/RFE to test things18:59
mtreinishclarkb: nope it does, it's kinda useful for testing regexes18:59
openstackgerritMerged openstack-infra/project-config: Run full tempest on centos7
*** akshai has joined #openstack-infra18:59
mtreinishespecially because I always screw them up19:00
reedclarkb, probably not needed since the basic use case I have in mind is of a new contributor who will not have to do the merge, right?19:00
openstackgerritEmilien Macchi proposed openstack-infra/project-config: puppet: run beaker on both trusty & centos7
clarkbreed: ya merge commits should only come from a project core/release manager19:02
clarkbreed: everyone else won't need to worry about them they will just push to the branch19:02
openstackgerritDoug Hellmann proposed openstack-infra/release-tools: update to set defaultbranch
*** spzala has joined #openstack-infra19:03
reedso, is storyboad the place to log such RFE?19:03
clarkbmtreinish: neat for some reason I didn't think list tests applied the regex and the regex was applied post listing19:04
clarkbreed: sure19:04
mtreinishclarkb: oh, that's the internal mechism19:04
mtreinishit does full discovery and then does a regex match afterwords19:04
*** patrickeast has quit IRC19:04
mtreinishor afterwards I guess :)19:04
openstackgerritMerged openstack-infra/project-config: Add testr_results report
reedi cannot see comments on storyboard19:06
reednow I can19:06
reedneeded to refresh19:06
reedok, I'm good to go to lunch19:06
*** dannywilson has quit IRC19:08
*** tsg_ has joined #openstack-infra19:09
*** shashankhegde has quit IRC19:10
*** dannywilson has joined #openstack-infra19:10
*** jamesmcarthur has joined #openstack-infra19:11
anteayasuperdan: thanks for you reply to lennyb19:12
*** ZZelle_ has joined #openstack-infra19:12
*** ZZelle_ has quit IRC19:12
superdananteaya: np19:13
*** hdd has quit IRC19:13
*** eharney has quit IRC19:15
*** tjones2 has joined #openstack-infra19:16
*** jamesmcarthur has quit IRC19:16
*** mrunge has quit IRC19:17
lifelessmordred: except python-quantumclient wasn't doing stable branches w/point releases :)19:19
lifelessmordred: still, I'll take your word :)19:19
*** ddieterly has quit IRC19:19
openstackgerritJan Klare proposed openstack-infra/project-config: move gate-.*-chef-rake job and run it branch specific
*** ddieterly has joined #openstack-infra19:21
*** rm_work is now known as rm_work|away19:22
lifelessjd__: unicode requirements? !?19:24
jd__lifeless: lol yeah I thought we were in 2015 but finally I gave up :)19:24
*** dustins has quit IRC19:24
lifelessjd__: I'm surprised 2.6 failed and 2.7 didn't19:25
openstackgerritEmilien Macchi proposed openstack-infra/project-config: Check more Puppet modules with tripleo-f20puppet-nonha
EmilienMdprince: ^19:25
lifelessjd__: well, surprised it failed *there*, vs higher up.19:25
jd__lifeless: I imagine has a different behaviour by default?19:25
lifelessjd__: FWIW I'd be entirely happy to make pbr str-bytes clean19:25
lifelessjd__: [and I do need to support 2.6 still, for the testing ecosystem]19:25
lifelessjd__: oh, I see what you did19:26
lifelessjd__: so in 2.6, the docs say The io module provides the Python interfaces to stream handling. The built-in open() function is defined in this module.19:27
pleia2cinerama: missed your comment earlier, the percentages thing may be a problem, since we only commit the translations to the repository w/ gerrit once they are a percentage completed, if Zanata really can't provide these statistics to the client (and we can't get them to write it in), we may have to write a script on the proposal slave to download everything and then determine percentages there19:27
mordredlifeless: how much longer do you need to support 2.6 in pbr?19:27
pleia2(oops, that line may have cut off)19:27
mordredlifeless: given that 2.6 has been EOL'd for a while ...19:28
lifelessmordred: indefinitely, but I'll happy provide the resources to do so19:28
lifelessmordred: 2.6 is still in widespread use19:28
mordredlifeless: k. I mean, I'm mostly just asking because infra will stop having 2.6 capable build hosts once openstack kills 2.6 support19:28
*** signed8bit_ZZZzz is now known as signed8bit19:28
cineramapleia2: the web UI provides the percentage completion so it's not like it's not available19:28
mordredso at some point we should figure out what support it looks like19:29
jd__lifeless: ah yeah so that makes sense that it fails :( I could try to specify an encoding but I don't think it's really good?19:29
lifelessmordred: so, 2.6 in a container is sufficient to test19:29
pleia2cinerama: yeah, so maybe a good thing to ask them about exposing via the client API?19:29
lifelessjd__: I think it will be a bit of a tarpit19:29
*** esker has quit IRC19:29
lifelessjd__: since setuptools internals and distutils internals we use are likely not hygenic yet19:29
lifelessjd__: so there'd be a chunk of pre-call conversions needed to avoid blowups in other code bases19:30
*** _dguerri is now known as dguerri19:30
lifelessmordred: you should look at
cineramapleia2: yup19:30
lifelessmordred: 2.6 is still ~15% of pypi downloads. I'm tempted to argue we should preserve client support for 2.6 generally on that basis19:31
*** teran has quit IRC19:31
lifelessmordred: since if folk are crazy enough to use it...19:31
*** teran has joined #openstack-infra19:31
pleia2cinerama: I *believe* that's the big one WRT downloads, off the top of my head I don't think there are any others19:31
lifelessjd__: how did you think to test this? was it a comment or an actual package with unicode in it ?19:32
*** akshai has quit IRC19:33
lifelessmordred: whats the limiting factor on infra having 2.6 capable hosts?19:33
jd__lifeless: it was a comment19:33
*** tjones1 has quit IRC19:33
mordredlifeless: on the other hand - perhaps EOL'ing client support on 2.6 might help convince people to move off it19:33
jd__lifeless: I don't think package with unicode is going to work :) I'm not that optimistic :)19:33
mordredlifeless: an OS that includes it19:33
mordredlifeless: that still gets security updates19:33
lifelessmordred: would be ok?19:33
mordredlifeless: we don't really run from PPAs except under extreme duress, no19:33
mordredand also - who is doing security fixes for 2.6 in that repo?19:33
clarkbnick wrote a blog on this iirc19:34
mordredI mean, at some point EOL has to mean EOL19:34
clarkband said please stop supporting python2.619:34
mordredand I'd like to think that three-years after EOL is good enough19:34
mordredand thing need to start breaking19:34
mordredso I agree with nick19:34
lifelessso nicks thing said19:34
lifelessif you're a project, stop caring, let redistributors care19:34
lifelessits their job19:34
mordredwe're a project19:34
mordredI think we do the community a service by ceasing to care19:35
mordredsometimes it takes outside help to get out of a bad relationship19:35
lifelessso unless a redistributor pops up to care19:35
*** teran has quit IRC19:35
mordredsure - but I can't do anything about that19:35
lifelessI'm going to be personally affected if pbr breaks on 2.6 before I EOL 2.6 caring in testtools/testscenarios/testr/subunit/linecache2/unittest2/traceback2/mock19:36
mordredlifeless: well, we tend to not land pbr changes quickly19:36
*** jtriley_ has joined #openstack-infra19:36
*** dhritishikhar has quit IRC19:36
lifelessmordred: I know, its more the ohfuckitbroke concerns around not being tested19:36
jtriley_hello all - just signed up for and having issues running git review -s19:36
mtreinishfungi, clarkb: is there an easy way to query which projects I have +2 on?19:37
*** isviridov_away is now known as isviridov19:37
jtriley_signed the ICLA and added my ssh key19:37
jtriley_and set my username in review.openstack.org19:37
lifelessmordred: so, whats the concern around security updates for a python thats only used to run a test suite as an unprivileged user ?19:37
openstackgerritElizabeth K. Joseph proposed openstack-infra/system-config: Add instruction to verify new server fingerprint
clarkbmtreinish yes, label:code-review,+2 iirc19:37
lifelessmordred: [putting aside the PPA aspect for a second]19:37
clarkbmtrienish oh wait19:37
clarkbmtrienish I misread you want to know if you can +2, I dont think so19:38
lifelessyou need to query the acls19:38
lifelessand team memberships19:38
lifelessIthink its accessible19:38
clarkbit is its not easy19:38
mtreinishlifeless: yeah, I was looking (through the ui at least) and it didn't seem like something exposed easily19:39
*** prad has quit IRC19:39
mtreinishbecause of the group heiarchies19:39
lifelessmtreinish:,access <- acls19:40
*** prad has joined #openstack-infra19:40
lifelessmtreinish: and from there to groups, and spider out from there.19:40
fungijtriley_: also signed up as an individual member of the openstack foundation and confirmed that supplying contact information in gerrit didn't return an error? and do you have (or did you set) a username in gerrit?19:40
fungijtriley_: confirm that the username field at matches what you're trying to use with git review19:41
greghaynesWhat is the ubuntu apt mirror we should be using during test jobs?19:41
mtreinishlifeless: yeah I realize that. I was hoping to do it in reverse though. Ie given a user find the access permissions on all projects19:41
lifelessif i wanted to make a variation of the pbr integration job that had a different pip version19:41
fungijtriley_: and that says something like "Contact information last updated on [some recent date and time]."19:41
lifelessmtreinish: so I suspect checking out the committed form of the acls is that19:41
mordredgreghaynes: we don't have one / set one19:42
*** tjones1 has joined #openstack-infra19:42
lifelessmtreinish: or I may be imaginging something19:42
mordredgreghaynes: it's been a thing we've discussed, but so far is not ugly enough of a problem that anyone has worked on it19:42
greghaynesI just noticed that the trove dib jobs are using archive.ubuntu.com19:42
lifelesswhats the right git repo to hack on these days to add a variation of a job19:42
lifelessinfra/system-config ?19:42
greghayneslifeless: jobs are in project-config19:43
*** rm_work|away is now known as rm_work19:43
*** ajo has joined #openstack-infra19:43
fungijtriley_: mainly double-check that you followed the steps at in sequence19:43
lifelessslow net is slow Receiving objects:  25% (13626/54502), 3.38 MiB | 94.00 KiB/s19:44
fungimordred: lifeless: i'm not _too_ concerned with the security posture of random debs used in running jobs as we're moving toward no longer preinstalling them anyway (just pre-caching them)19:45
fungiand also not talking about installing them on long-running systems, production servers, whatever19:46
mordredpatch bomb incoming19:46
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a method for getting an endpoint
openstackgerritMonty Taylor proposed openstack-infra/shade: Make keystone endpoints work like the others
openstackgerritMonty Taylor proposed openstack-infra/shade: Add early service fail and active check method
openstackgerritMonty Taylor proposed openstack-infra/shade: Call super in OpenStackCloudException
clarkbI have somehow forgotten that I should get a thermometer for the smoker19:46
mordredclarkb: FAIL19:46
* clarkb runs off to find lunch and kitchen gadgets19:46
greghaynesclarkb: :O19:46
mordredgreghaynes, SpamapS: ^^ I haz fixed19:47
greghaynesclarkb: I have my thermometer for brewing19:47
fungithe only real security concern there i guess is that by saying "commit tested against package <whatever>" could make downstream consumers think we're endorsing the security of that package19:47
greghaynesclarkb: I also have a thermocouple we can hack on :p19:47
clarkbgreghaynes: will you be awake at 5am tomorrow? I guess hacking needs to happen before that19:47
clarkbmeh I will just buy a remote probe thermometer19:48
*** eharney has joined #openstack-infra19:48
fungithey're cheap these days19:48
morganfainbergclarkb, fungi, jeblair, pleia2: I'd like to get a branch created for pycadf for stable/juno19:48
morganfainbergthe commit it should be based on is:
morganfainbergcc ttx ^19:49
morganfainbergthis is to fix keystonemiddleware's failing19:49
*** dangers_away is now known as dangers19:49
fungidhellmann: ^ do you or other release managers want to be doing that or okay with one of us doing it?19:49
morganfainberggordc, bkudson, ^^19:49
* fungi hopes ttx has checked out for the weekend19:49
morganfainbergfungi, seriously i do to19:49
lifelessoh man19:49
lifelessexport PIP_PROCESS_DEPENDENCY_LINKS=119:49
lifelesswe so have to kill that19:49
lifelessin our tests19:50
fungimorganfainberg: i can do it easily enough, but let's give it a few minutes for someone to respond. i don't want to step on any release management toes19:50
morganfainbergfungi, thanks.19:50
morganfainbergfungi, totally.19:50
morganfainbergfungi, :)19:50
lifelessclarkb: I need a small hint for jjb19:50
lifelessclarkb: I want to add a variation of check-pbr-installation-dsvm19:50
lifelessclarkb: that uses a different pip version19:50
bknudsononce that's created we'll probably want the proposal bot to update its requirements19:51
lifelessclarkb: do I encode that into the name19:51
dhellmannfungi: I'm OK with you doing it, but I also have time to help morganfainberg if you're busy with other things19:51
bknudsonand then we can also see if the tests work.19:51
lifelessclarkb: like ...-upstream in the layout and then pull that out with a {} variable in requirements.yaml ?19:51
*** isviridov is now known as isviridov_away19:51
dhellmannfungi, morganfainberg: you need to create the branch based on a release tag, right?19:52
bknudson0.6.0 is commit 52727bcea3...19:52
jtriley_followed all that as far as i can tell - meh i'll try again later19:52
gordcdhellmann: yep. it'd be off 0.6.0
morganfainbergdhellmann, ^19:52
dhellmanngordc, morganfainberg : ok, just double checking19:53
fungidhellmann: please take care of it, if you don't mind. it's less that i don't have time, but more that i like the release managers handling this for official projects19:53
dhellmannfungi: sure thing19:53
clarkblifeless ya needs to be a new name but likely you can use existing template19:53
fungidhellmann: and thanks!19:53
dhellmanngordc, morganfainberg : so this is stable/juno at 0.6.0, right?19:54
clarkblifeless but update the job in the template to switch on the parameter passed to the name19:54
*** Bsony has quit IRC19:54
*** nelsnelson has quit IRC19:54
gordci guess we should also add a disclaimer the 0.6.1 release would be a single patch which moves a oslo.messaging req from requirements.txt to test-requirements.txt?19:54
morganfainbergdhellmann: I defer to gordc and bknudson for specifics here. thye've been fighting this.19:54
gordcdhellmann: yep19:54
lifelessclarkb: tell you what, I'll blindly do it and you tell me in review, I tink thats most efficient for all concerned19:54
dhellmanngordc: a requirements change is really a semver minimum version bump :-/19:54
fungijtriley_: can you paste the error you're getting at and link the resulting url here? it might help us tell you where the problem is19:54
dhellmanngordc, morganfainberg : let's move to openstack-relmgr-office to cut down on the noise here19:54
dhellmannbknudson: ^^19:55
mordredlifeless: what's doing PIP_PROCESS_DEPENDENCY_LINKS?19:55
jtriley_fungi: sure it's git-review telling me could not connect to gerrit and asking for my username19:55
lifelessmordred: grep for it in project-config19:55
jtriley_not much to see19:55
sc`dumb question: jenkins is supposed to be automatically added as a reviewer to every review, but it's not on a change i recently submitted. should that be of concern?19:55
*** isviridov_away is now known as isviridov19:56
*** HeOS has joined #openstack-infra19:56
jtriley_fungi: Trying again with ssh:// 'we dont know where your gerrit is'19:56
fungijtriley_: that can sometimes be a misleading error indicating that the test push it tried failed for other reasons. what ssh username are you connecting with?19:56
jtriley_same as my irc (without the _)19:56
fungiah, i see it above19:57
jtriley_i cloned the repo from github which the project owner does as well apparently19:57
pleia2grabbing some lunch, bbiab19:57
*** tsg_ has quit IRC19:57
fungijtriley_: i see that username corresponds to account_id 16110 in gerrit. can you confirm that's the account id number you see at
jtriley_i see it19:58
openstackgerritMerged openstack-infra/project-config: Add python34 unit test job to tempest experiemental
*** tnovacik has joined #openstack-infra19:58
*** patrickeast has joined #openstack-infra19:59
fungijtriley_: it looks like that account has not supplied contact information. go to and let me know if you see "Contact information last updated on <some date and time>."19:59
fungijtriley_: normally you're prompted for that when you agree to the icla, but might have missed entering it for some reason20:00
jtriley_ok i generally avoid that type of info if it will let me ;)20:00
jtriley_which i guess it did..20:00
fungijtriley_: go ahead and put something (anything really) in the mailing address and country fields there and save it20:00
openstackgerritlifeless proposed openstack-infra/project-config: Test pbr against unreleased pip.
openstackgerritAurélien Chabot proposed openstack-infra/git-review: Add an option to add reviewer to submitted changes
jtriley_that did it!20:01
jtriley_fungi: thank you very much20:01
*** tjones2 has quit IRC20:01
fungijtriley_: what's probably happening is that git-review is being told during its test push that the kolla repository's cla requires contact information, but git-review is hiding that error and assuming your username is wrong. i believe it's fixed in git-review master but not released yet20:01
*** otter768 has joined #openstack-infra20:02
*** marcusvrn1 has quit IRC20:02
jtriley_ah ok good to know thanks again for your help20:02
*** tsg_ has joined #openstack-infra20:02
fungijtriley_: absolutely--any time! happy hacking20:02
*** ociuhandu has quit IRC20:02
sdakefungi as usual #openstack-infra is a wealth of knowlege :) tia :)20:03
fungisdake: it's only because we beat our heads against these particular problems until they're imprinted into our skulls20:03
sdakefungi yup I know :)20:03
sdakefungi I see it all day long when I lurk in this channel :)20:04
fungitomorrow we'll fix everything, i promise. it's too bad tomorrow never comes20:04
sdakefungi there is always tomororw :)20:04
sdake(I use that on my kids all the time :)20:05
sdake"daddy I'll do it tomorrow"20:05
sdake"no you will do it today!"20:05
fungione of the bars i frequent has a big sign on the wall that says "free beer tomorrow"20:05
*** otter768 has quit IRC20:06
*** kgiusti has left #openstack-infra20:08
lifelessclarkb: ^ thoughts appreciated20:09
*** jtriley_ has quit IRC20:09
*** haypo has joined #openstack-infra20:09
fungisc`: which review? got a link?20:09
sc`fungi: looks like something knocked loose. after about 10 minutes it showed up20:10
fungisc`: jenkins won't show up as a reviewer until the first time it finishes testing the review and leaves a comment/vote on it20:11
sc`got it. still getting my head wrapped around how it all works20:12
fungisc`: sometimes it can take a few minutes to run test jobs (or even to process the upload event if it's particularly busy)20:12
fungilifeless: i suspect one or two of the jobs are going to barf on the stray hard tab in your yaml20:13
*** jgrimm is now known as zz_jgrimm20:14
*** akshai has joined #openstack-infra20:15
haypohi. i don't understand why gate-tempest-dsvm-largeops-src-oslo.messaging fails with : my change on requirements.txt should not impact the installed eventlet version: -- i saw these tracebacks:
openstackLaunchpad bug 1403284 in devstack "mysql (1040, 'Too many connections')" [Undecided,In progress] - Assigned to Joe Gordon (jogo)20:15
fungilifeless: curious, have you tested running devstack-gate with PBR_PIP_VERSION exported like that in the calling environment?20:16
haypoit looks like "Bug 1403284 - mysql (1040, 'Too many connections')" only occured on my changeset:
lifelessfungi: narf, thanks20:17
openstackgerritlifeless proposed openstack-infra/project-config: Test pbr against unreleased pip.
lifelessfungi: I'm adding the glue to now20:17
*** rm_work is now known as rm_work|away20:17
*** Bsony has joined #openstack-infra20:17
fungilifeless: also this could be done by extending the existing job-template similar to how we do the {branch-designator} parameter suffix on other d-g jobs, but what you have there isn't technically wrong20:17
lifelessfungi: yeah, I couldn't see an easy cargo cult20:18
lifelessfungi: so wanted a reviewer to say 'it should look like <details>'20:18
fungiit's not crucial in my opinion20:18
lifelessfungi: I'm not happy with what I have, is all I'm saying20:18
fungibut i can add notes if you're interested in trying to keep one template there20:18
lifelessthere are currently two with the only difference being s/requirements/pbr/ in one command AFAICT20:19
lifelessthat offends me :). making it be just one would be great.20:19
lifelessmy patch adds a third20:19
openstackgerritMerged openstack-infra/project-config: Add qemu-img to dib functests
*** rm_work|away is now known as rm_work20:19
lifelessbut I am perhaps overly nuts on this20:19
lifelesscertainly I'd like to keep the two pbr specific ones unified20:19
fungilifeless: agreed, reduplicating nearly identical jobs is what templates were intended to solve ;)20:19
*** woodster_ has quit IRC20:20
*** mrmartin has quit IRC20:21
*** ibiris is now known as ibiris_away20:22
*** Bsony has quit IRC20:23
*** shashankhegde has joined #openstack-infra20:25
lifelessI think is broken20:25
*** hashar has quit IRC20:25
lifelessI need to check a couple thing20:26
fungithat wouldn't _entirely_ surprise me20:26
lifelessI think it runs20:26
lifelessbut I'm not sure that it populates the wheelhouse it thinks it is20:26
fungialso remember that script has been forked and has different versions in pbr vs requirements repos now20:26
lifelessoh gawd20:26
lifelesswhy can't we have nice things?20:26
*** Bsony has joined #openstack-infra20:27
lifelessyeah WHEELHOUSE isn't a pip variable20:27
lifelesswe're not writing our wheels to the wheelhouse we've made20:27
lifelesswe're writing them to ./wheelhouse20:27
fungithat could certainly impact performance20:28
lifelessthis first20:28
mordredlifeless: I look forward to it sucking less20:28
lifelessthen that20:28
*** tiswanso has quit IRC20:29
*** Swami has quit IRC20:29
lifelessyeah, these things need to be options to pip20:29
fungithough some of the challenge is that passing pip options through tox without modifying tox.ini is tough20:30
fungii guess pip.conf is the other answer there20:30
lifelessthis is the integration tests though20:30
fungiahh, yeah in that case i don't think we're relying on tox20:30
lifelesswe're not going to run project tests20:31
lifelesswe're going to run tempest20:31
lifelesshaving put everything together with a new pbr || new pip20:31
*** peristeri has quit IRC20:31
lifelessor maybe its just referring to using the dsvm image20:31
*** Bsony has quit IRC20:32
mordredyeah - it just uses the image20:33
mordredSpamapS: python2.7 does NOT like my changes to the Exception20:34
*** teran has joined #openstack-infra20:35
mordredoh. hah20:35
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a method for getting an endpoint
openstackgerritMonty Taylor proposed openstack-infra/shade: Make keystone endpoints work like the others
openstackgerritMonty Taylor proposed openstack-infra/shade: Add early service fail and active check method
openstackgerritMonty Taylor proposed openstack-infra/shade: Call super in OpenStackCloudException
* mordred is fascinated as to why that did not break python3.420:36
lifelessalso the next pip release will build wheels itself20:36
lifelessso all the manual wheel caching can be tossed20:36
mordredlifeless: \o/20:37
*** dangers is now known as dangers_away20:37
*** dangers_away is now known as dangers20:38
fungilifeless: some comments added to 177423 on extending the current template20:38
fungiand also on how to fix the current job failure20:38
*** esker has joined #openstack-infra20:40
*** woodster_ has joined #openstack-infra20:40
lifelessmordred: indeed :)20:40
lifelessmordred: yeah, I did that :)20:41
* mordred hands lifeless a used packet of seasoning20:43
lifelessfungi: why requirements-jobs ?20:43
mordredlifeless: because20:43
SpamapSmordred: oh?20:43
lifelessfungi: I don't want to test that a change to global-requirements works with upsream tests20:43
fungilifeless: it's a pseudo project name20:44
mordredSpamapS: enjoy the diff of diffs between the last two patchsets of 17734720:44
fungilifeless: it can be renamed to gralpy if we want20:44
fungiin fact i'd posit that gralpy would be better anyway20:44
*** mattfarina has quit IRC20:45
*** isviridov is now known as isviridov_away20:45
SpamapSmordred: thats exactly what I was just looking at20:45
SpamapSmordred: why do you need to build up real_args ?20:45
openstackgerritlifeless proposed openstack-infra/project-config: Test pbr against unreleased pip.
lifelessfungi: I don't know what that means, but doing what you say anyway20:45
SpamapSmordred: why not just pass the two that you expect to super?20:45
fungilifeless: about the only thing project entries do in jjb is provide a convenient mechanism to fill in the {name} built-in parameter20:46
fungilifeless: particularly so as to pass them into job-templates you're instantiating20:46
mordredSpamapS: I dunno20:46
fungilifeless: for example there's another one in there called devstack-jobs. it's not a real git repo either, just another dummy project name used for collecting job-template instantiations20:47
*** rm_work is now known as rm_work|away20:47
openstackgerritMonty Taylor proposed openstack-infra/shade: Call super in OpenStackCloudException
*** rm_work|away is now known as rm_work20:47
mordredSpamapS: like that?20:47
SpamapSmordred: yes like that. :-D20:48
mordredk. I push rest of stack20:48
SpamapSmordred: assuming that actually works... :)20:49
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a method for getting an endpoint
openstackgerritMonty Taylor proposed openstack-infra/shade: Make keystone endpoints work like the others
openstackgerritMonty Taylor proposed openstack-infra/shade: Add early service fail and active check method
mordredSpamapS: only one way to find out20:49
*** samuelBartel has quit IRC20:49
lifelessmordred: why are you calling super ?20:50
lifelessmordred: also, why map(repr) instead of %r in the format ?20:52
*** pelix has quit IRC20:52
*** BobH1 has quit IRC20:53
fungilifeless: much closer, but my explanation about how to add what you need to the jenkins/jobs/projects.yaml was pretty lacking, sorry. clarifications in the latest patchset20:54
*** spzala has quit IRC20:56
*** BobH has quit IRC20:57
*** rm_work is now known as rm_work|away20:57
*** dguerri is now known as _dguerri20:57
*** armax has quit IRC20:57
bknudsonso we created a branch stable/juno for pycadf... is proposal bot going to come along and update requirements automatically?20:58
*** patrickeast has quit IRC20:58
mordredlifeless: %r looks like it would add [] around the args, which would not be happymaking20:58
fungibknudson: the next time a change merges to the stable/juno branch of the openstack/requirements repo it should20:59
mordredlifeless: and calling super because I'm only adding one param myself, and the other one should be have as standard Exception does20:59
bknudsonfungi: that might never happen.20:59
*** tnovacik has quit IRC21:00
fungibknudson: in the meantime, if you're in a hurry, you could run the script from requirements against that branch of pycadf and propose the resulting diff yourself (after checking that it makes sense of course)21:00
mordredSpamapS: if I stab you in the face in a friendly way, don't take it the wrong way21:00
fungibknudson: anything else in that particular sausage factory you have questions about, don't hesitate to ask21:01
*** Swami has joined #openstack-infra21:01
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a method for getting an endpoint
openstackgerritMonty Taylor proposed openstack-infra/shade: Make keystone endpoints work like the others
openstackgerritMonty Taylor proposed openstack-infra/shade: Add early service fail and active check method
openstackgerritMonty Taylor proposed openstack-infra/shade: Call super in OpenStackCloudException
SpamapSmordred: my face hasn't been stabbed in a while. ;)21:02
* mordred uses all of the test nodes ...21:02
mordredSpamapS: ok. I think that's finally a winner21:02
anteayamordred: have you been running low on cats?21:02
*** gyee has quit IRC21:02
fungimordred: you're just stress-testing hpcloud, right?21:02
mordredanteaya: I try to use them only on special occasions21:02
mordredfungi: yup21:02
*** gyee has joined #openstack-infra21:02
anteayamordred: I feel that is such a nicer form of communication21:02
* anteaya feels bad about the stabbing21:03
mordredanteaya: I did hand lifeless a used packet of seasoning earlier21:03
*** tsg_ has quit IRC21:03
*** bswartz has quit IRC21:03
anteayaI missed that21:03
mordredfungi: nope21:03
mordredfungi: irc21:03
fungioh, too bad21:03
anteayasuch I nice gift21:03
fungi'cause i totally pictured it happening in meatspace21:03
anteayaha ha ha21:04
*** ayoung has joined #openstack-infra21:04
*** isviridov_away is now known as isviridov21:04
*** sdake_ has joined #openstack-infra21:04
*** tsg has joined #openstack-infra21:04
openstackgerritlifeless proposed openstack-infra/project-config: Test pbr against unreleased pip.
lifelessfungi: thanks21:05
lifelessfungi: I appreciate the hand holding immensely21:05
*** dhritishikhar has joined #openstack-infra21:05
mordredfungi: now I kind of want to hand lifeless a used packet of seasoning in person21:05
fungilifeless: other than the hard tabs your editor seems to really like inserting in yaml, i think that should pass our project-config jobs now21:06
lifelessfungi: so frustrating21:06
mordredmy editor is the WORST at yaml, btw21:06
*** andreykurilin__ has joined #openstack-infra21:06
fungiis it like worstcat for yaml?21:07
mordredit constantly wants to reindent lines in the wrong way21:07
mordredfungi: YES21:07
openstackgerritlifeless proposed openstack-infra/project-config: Test pbr against unreleased pip.
lifelessI need to add a yaml rule for my vimrc21:07
*** salv-orlando has joined #openstack-infra21:07
mordredlifeless: if you find good vim settings to make yaml not suck, please share21:07
mordredlifeless: I don't have tab problems, but I do have bad autoindent problems21:08
*** sdake has quit IRC21:08
lifelessmordred: I'll fiddle when I have a spare cycle21:08
* mordred hands lifeless a bicycle he found out on the street, ignoring the half-eaten sandwich left in the basket21:09
fungilifeless: lgtm. now it's clarkb's turn to tell me i've steered you in an entirely wrong direction or something21:09
SpamapSmordred: that cat sandwich looks disgusting.21:09
cineramapleia2: oh SNAP. we're down. can you look?21:10
*** jtriley has joined #openstack-infra21:10
*** tonytan4ever has quit IRC21:10
fungicinerama: i can take a look if she's convalescing with her plague at the moment21:10
fungicinerama: translate-dev.o.o yeah?21:11
cineramafungi: lemme put on my plague doctor mask21:11
cineramafungi: yes21:11
fungiapache's running at least21:11
fungicinerama: i can get to the login screen at least21:12
fungiwhat specifically is broken?21:12
*** salv-orl_ has joined #openstack-infra21:13
cineramafungi: oh weird, now it's working. before it was just showing the wildfly default screen. i noticed because the client was having a sad21:13
cineramafungi: it might be worth seeing if there's something of interest in /var/log/wildfly/console.log21:13
fungii'll take a peek. might be cats in the gears21:13
openstackgerritlifeless proposed openstack-dev/pbr: Honour new test variable PIPVERSION
*** salv-orlando has quit IRC21:14
*** dboik_ has joined #openstack-infra21:15
lifelessfungi: $tmpdir/wheelhouse/bin/pip21:15
*** akshai has quit IRC21:15
fungicinerama: the earliest timestamp in that file is 2015-04-24 21:09:5821:15
fungicinerama: i smell a restart21:15
yolandamordred, jeblair, would you mind taking a look at ? it's becoming a bit urgent from our side, and already has a +2 and several positive comments21:15
yolandanodepool image testing21:15
fungilifeless: what's the question?21:16
lifelessfungi: Can lifeless copy-paste correctly.21:16
*** ldnunes has quit IRC21:16
lifelessoh, I see it21:17
openstackgerritlifeless proposed openstack-infra/project-config: Test pbr against unreleased pip.
mordredyolanda: I think it needs a jeblair21:18
anteayajust one jeblair?21:18
*** dboik has quit IRC21:18
mordredanteaya: how many jeblairs does it take?21:19
anteayausually less than one21:19
*** sdake has joined #openstack-infra21:19
*** dboik_ has quit IRC21:19
anteayawas picturing many jeblairs21:19
anteayathe libraries they would organize21:19
mordredis a jeblair a metric or imperial unit?21:19
anteayaI would put money on metric myself21:20
*** salv-orl_ has quit IRC21:20
yolandamordred, or any other infra cores , any of you guys are good to review that :)21:20
fungicinerama: so it appears that wildfly restarted, though the server itself seems fine. also the rotated log doesn't seem to contain any indication of why it went down/restarted21:20
fungicinerama: no sign of oom condition in dmesg or anything21:20
fungicinerama: puppet applied a new configuration right then and restarted it21:21
yolandamordred, i just need to start moving that a bit, even if that's not approved but i have good feedback on the direction21:21
cineramafungi: wondering if puppet ran again21:21
*** ajo has quit IRC21:21
fungicinerama: mystery solved21:21
fungiApr 24 21:10:22 translate-dev puppet-agent[14575]: (/Stage[main]/Wildfly::Service/Service[wildfly]) Triggered 'refresh' from 1 events21:21
cineramafungi: great, now we can all hang out with the gang at $local_establishment and end on a freeze frame21:21
*** haypo has quit IRC21:22
fungicinerama: it would have gotten away with it too, if it weren't for us meddling kids21:22
openstackgerritBrant Knudson proposed openstack/requirements: Bandit should not be updated automatically
*** sdake_ has quit IRC21:23
*** viglesias has quit IRC21:25
*** ajo has joined #openstack-infra21:25
*** e0ne has joined #openstack-infra21:27
*** e0ne has quit IRC21:29
*** viglesias has joined #openstack-infra21:31
lifelessfungi: 177423 is green21:31
lifelessmordred: / clarkb: ^ could I get a +A ?21:31
*** e0ne has joined #openstack-infra21:33
clarkbyolanda reviewed the spec21:34
*** ChuckC_ is now known as ChuckC21:34
yolandaclarkb, thx, let me take a look21:35
*** radez is now known as radez_g0n321:36
yolandaclarkb, what do you mean with I don't know that nodepool should rerun tests multiple times. You can just have the test itself retry if that is desired.21:36
clarkblifeless you added a periodic job to jenkins but zuul layout doesnt seem to run it. Is that intentional?21:36
yolandamy idea was to add a retry setting,that would make nodepool retry the test if failed, the desired amount of times21:37
clarkbyolanda put the for loop in the script that runs on the node not in nodepool21:37
yolandaclarkb, so you mean add the number of retries in the same script is going to test the image?21:37
clarkbyolanda in my mind the interface here should be boot need, run script, check return code21:37
clarkbits simple and matches existing interfaces21:37
clarkbwe already do this for image builds and node ready checks21:38
yolandaok, i'm fine with that. And for running the test, you should just spin up the node, and run the test as part of that spin up ? you meant that'21:39
clarkbya basically do like the ready check script but for an image state transition21:39
lifelessclarkb: my intent was to run it.21:39
yolandai'm ok with your opinion, even much more simple than my thought21:39
clarkbI dont like making jenkins and gearman required here since neither are required today21:39
lifelessclarkb: I am just entirely paged out all of this21:39
*** signed8bit has quit IRC21:40
yolandai'll just amend it now and push, as I want to move that forward as soon as possible21:40
fungiclarkb: good point, i missed that should have been periodic-whatever in the periodic pipeline21:41
clarkbit should allow you to optionally use jenkins or anything else21:41
*** akshai has joined #openstack-infra21:41
clarkblifeless fungi I have to drive noe but lgtm other than not running the periodic job21:41
*** dangers is now known as dangers_away21:41
lifelessmorded has +A21:42
lifelessI will follow up with the periodic thing21:42
*** dangers_away is now known as dangers21:42
lifelesszuul/layout you say21:42
fungilifeless: yeah, so the periodic instantiation of that job-template would simply allow you to have a different job name for the version you put in the periodic pipeline in layout.yaml21:43
*** miqui has quit IRC21:43
lifelessfungi: oh, so it will run periodically now?21:44
fungilifeless: yeah, i don't think that extra stanza in projects.yaml is actually needed21:44
*** gordc has quit IRC21:44
lifelessthen we'll wait and see21:45
openstackgerritMerged openstack-infra/project-config: Test pbr against unreleased pip.
fungilifeless: we have some jobs where the lack of a git commit to test on is a problem and so there are separate periodic jobs/job-templates to deal with that situation21:45
fungilifeless: but without having thought through this particularly well, i suspect the pbr integration test job won't care21:45
lifelesshow long till that new commit is active?21:46
fungiit's _possible_ devstack-gate will just get all unhappy, but we can wait and see i suppose21:46
lifelessI need to recheck the new change in pbr against it21:46
*** hdd has joined #openstack-infra21:46
fungilifeless: best guess is ~30-45 minutes before the jobs are created on the jenkins servers and zuul has reconfigured to know to run them21:46
lifelessI'll pop back in a bit then21:46
*** mriedem has quit IRC21:47
*** bknudson has quit IRC21:49
*** smccully has quit IRC21:52
*** stevemar has quit IRC21:53
*** dhritishikhar has quit IRC21:53
*** dboik has joined #openstack-infra21:54
*** prad has quit IRC21:57
*** prad has joined #openstack-infra21:58
*** e0ne has quit IRC22:00
*** baoli has quit IRC22:00
openstackgerrityolanda.robla proposed openstack-infra/infra-specs: Add nodepool image testing spec.
yolandaclarkb, i pushed a new version. Many thx for your feedback22:02
*** jtriley has quit IRC22:03
*** otter768 has joined #openstack-infra22:03
*** mwagner_lap has quit IRC22:03
*** hdd has quit IRC22:04
*** jtriley has joined #openstack-infra22:06
*** otter768 has quit IRC22:07
*** yamamoto has joined #openstack-infra22:08
*** patrickeast has joined #openstack-infra22:09
*** isviridov is now known as isviridov_away22:11
*** sdake_ has joined #openstack-infra22:11
*** Kennan2 has joined #openstack-infra22:12
*** Kennan has quit IRC22:12
*** _dguerri is now known as dguerri22:12
*** yamamoto has quit IRC22:13
*** lascii is now known as alaski22:13
*** r-daneel has quit IRC22:15
*** sdake has quit IRC22:15
*** bswartz has joined #openstack-infra22:15
clarkbfungi: re 177044 why would project bootstrappers have any explicit perms? They can basically do anything already22:16
openstackgerritDavide Guerri proposed openstack-infra/shade: WiP: Add Neutron/Nova Floating IP support
clarkbfungi: does exclusive group perms override the magical do everything permissions for project boostrappers?22:17
fungiclarkb: exclusive group permissions takes it away if not explicitly mentioned in that section22:17
fungiwe might actually consider dropping that in per-project acls though anyway, just because administrators can work around it by temporarily adding themselves to whatever group has access to what they need22:18
*** teran has quit IRC22:18
clarkbfungi: I just tested and that doesn't appear to be the case22:18
*** akshai has quit IRC22:18
clarkbadd yourself to project-bootstrappers then look at,publish22:18
yolandamordred, clarkb, how is status will the multi-threaded nodepool change i did? do you think it won't be useful now that hpcloud is performing better? i was still interested in giving that a try22:18
clarkbthats a stable/kilo barbican change22:18
clarkbyolanda: it may still be useful, we had to pull it out of production because we think it may have caused hpcloud to fall over (though I was afk for that so may have misunderstood)22:19
fungifor example, the other day i needed to manually submit a change on devstack stable/juno and had to manually add myself to devstack-core to do it. just adding myself to project bootstrappers was insufficient22:19
yolandaoh really? my two queue change made hpcloud fail?22:19
clarkbfungi: oh nevermind its only code-review that is currently affected22:19
clarkbyolanda: all we have is the inference that we put it in place and a few hours later hpcloud said turn it off22:20
clarkbyolanda: same thing happened when we tried jeblairs change22:20
yolandaoh man...22:20
pleia2cinerama: ah, we'll just have to keep an eye on what triggers restarts (since they take a couple minutes and can be disruptive), some of the updates can probably pile up until we have a maintenance window in production22:20
clarkbyolanda: you probably need to talk to mordred about it bceause I don't know much more than that22:21
yolandaclarkb, so i had it running for 2 days on my cloud account, and was performing so good . But of course you have much more volume than me22:21
yolandamordred, do you know more?22:21
mordredno - I do not22:21
clarkbmordred: we put in yolanda's two queue change for nodepool so that list nodes would be separated and we very quickly disabled hpcloud after22:21
clarkbmordred: this was the first time we disabled hpcloud recently and I think they told you to do it22:22
*** dguerri is now known as _dguerri22:22
yolandaclarkb, i thought it was due to maintenance'22:22
mordredoh - yeah - I'm not sure I realized we'd put in yolanda's change22:22
openstackgerritJoe Gordon proposed openstack-infra/devstack-gate: Remove heat from default configuration
*** jtriley has quit IRC22:22
mordredso I think it might have just been happenstance22:22
mordredtheir system was just steadily degrading and hit the point where our load was too much22:22
clarkbfungi: I just approved the project bootstrappers cleanup22:22
yolandaso i wanted to give that a try but i don't want to make hpcloud explode22:22
mordredthankfully that's fixed22:22
yolandamordred, are you giving that patch another try?22:23
clarkbmordred: it may be, except the time before last that we did similarto nodepool we also had to turn it off22:23
mordredclarkb: nod22:23
*** sdake has joined #openstack-infra22:23
fungiclarkb: thanks. that was my bad for not noticing the normalize script was insane and had done that to the acls when i ran it before submitting the stable/kilo overrides before22:23
clarkbbut maybe everything is happy again post cleanup22:23
mordredclarkb: it may be worth trying again post cleanup22:23
clarkbmordred: ya22:23
mordredotoh - the patch shouldn't be needed in a world where the cloud isn't broken in the first pace22:23
yolandamordred, i know, it's not so broken now, but still a bit slow22:24
openstackgerritMerged openstack-infra/project-config: Add coverage job to cue check pipeline
clarkbyolanda: also I think jeblair had comments, have those been addressed?22:24
yolandalooking at numbers it still makes a bit of sense22:24
clarkbmaybe its best to try it again before rewriting it if possible22:24
yolandaclarkb, i think so, unless i missed something new22:24
clarkbyolanda: it was just the task interface stuff22:25
clarkbwhich isn't that new22:25
yolandaah, the stop22:25
yolandai did the change22:25
*** hyakuhei has joined #openstack-infra22:25
*** rlandy has quit IRC22:25
*** yamahata has joined #openstack-infra22:25
*** dimtruck is now known as zz_dimtruck22:25
yolandaclarkb, mordred, so i added a stop method to the taskmanager22:26
*** sdake_ has quit IRC22:27
mordredyolanda: is that a change that shade would need to support?22:27
clarkbthat was the concern with the first pass22:27
*** dprince has quit IRC22:28
yolandamordred, as far as i know, shade will allow a taskmanager to be passed22:28
yolandaand all the updates are done into taskmanager, i'm not doing changes on Task anymore22:28
openstackgerritMerged openstack-infra/project-config: There is no Project Bootstrappers-core...
*** jtriley has joined #openstack-infra22:30
clarkbanyone else want to review devstack-gate's foray into using ansible?
clarkbfungi: pleia2 jhesketh SergeyLukjanov ^ otherwise I am going to approve (it has necessary +2s)22:32
lifelessfolk interested in pip/pbr - see
lifelessfor automatic wheel building at work22:32
clarkblifeless: any idea why its doing `pip wheel ''` ?22:32
clarkblifeless: it seems to barf on that22:32
* clarkb is guesing a bad bash for loop?22:33
pleia2clarkb: having a look22:33
openstackgerritMerged openstack-infra/project-config: zuul/layout: make gate-puppet-.*-puppet-syntax-future voting
clarkbpleia2: danke22:34
openstackgerritJoe Gordon proposed openstack-infra/project-config: Add tempest-dsvm-sahara
openstackgerritJoe Gordon proposed openstack-infra/project-config: Add sahara tempest job to sahara and tempest
openstackgerritMonty Taylor proposed openstack-infra/shade: Add inventory command to shade
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
openstackgerritMonty Taylor proposed openstack-infra/shade: Add a method for getting an endpoint
openstackgerritMonty Taylor proposed openstack-infra/shade: Make keystone endpoints work like the others
openstackgerritMonty Taylor proposed openstack-infra/shade: Add early service fail and active check method
openstackgerritMonty Taylor proposed openstack-infra/shade: Split exceptions into their own file
mordredSpamapS, Shrews: ^^ I split out exceptions into their own file22:36
clarkbhrm jd__ is likely out for the weekend /me approves and leaves a comment22:37
yolandaclarkb, mordred, leaving for today, just drop some comments here if you have some more feedback about this change. I really wish i could talk to you more often, spending some time in the chat at your timezone is so useful22:37
*** thinrichs has joined #openstack-infra22:37
yolandai raised the comment with jeblair this week, communication with upstream is one of my main concerns22:38
mordredyolanda: ++22:38
mordredyolanda: I totally agree - it's hard when the overlap is bad22:38
clarkbsilly timezones22:38
yolandai will end up working at nights :)22:38
clarkbI tell you what though, it was a lot easier for me to stay up late and work with apac before I started having all these scheduled appointments during "normal" work hours22:39
clarkbunfortunately if you are seeing the doctor at 10pm thats usuall a very bad thing22:39
mordredclarkb: if you get a sec - 177476 and the two before it are boring and I'd love to clear them out of the queue - the ones after it I need to add some tests for ... but I'm not adding unit tests to test str(e) ...22:40
*** ayoung is now known as ayoung-noms22:40
yolandaso today i'm just talking with you because i'm on GMT time , an hour later than in Spain, and not with family. But it's still a crazy hour, so imagine in a day to day in spanish time, is so hard to have a chat with you all22:40
*** harlowja_ has joined #openstack-infra22:41
openstackgerritEmilien Macchi proposed openstack-infra/project-config: puppet: run beaker on both trusty & centos7
mordredyolanda: yah - it's one of the reasons I want more folks in europe timezones so that there are more people working together - but also since it's so hard to talk I fear growing two split teams ... it's a hard problem we have to solve22:41
*** erlon has quit IRC22:41
*** harlowja has quit IRC22:41
yolandamordred, so you have some folks interested on that, including myself, we'd love to resolve that situation22:42
lifelesshmm no, that hasn't stuck quite right22:42
yolandanot sure on how that works on other projects, i suppose it should be the same problem everywhere?22:42
fungilifeless: have an example run you can link?22:43
*** Swami_ has joined #openstack-infra22:43
thinrichsHi all.  If someone has a free minute, I'm having some trouble pushing tags/branches to gerrit (for the first time).  Possibly something silly.22:43
*** esker has quit IRC22:43
*** amitgandhinz has quit IRC22:44
fungithinrichs: can you put a transcript of the commands/errors on and link them here?22:44
thinrichsfungi: sure.22:44
lifelessfungi: I'm looking in for PBR_PIP_VERSION22:44
lifelessfungi: which should show up in the +if [ -n "${PBR_PIP_VERSION:-}" ]; then22:45
lifelesswhich does pip wheel -w22:45
lifelessbut there's no pip wheel -w22:45
*** rbradfor has quit IRC22:45
mordredyolanda: I think we're a slightly different situation, because we're a little more protective of the core status due to the production systems hanging off of the +2 ... that doesn't mean we're doing it right currently - but I think it's at least slightly harder for us to solve than for the other teams22:46
thinrichsfungi: I'm a member of the congress-release on gerrit22:46
lifelessfungi: and we should be installing a pip wheel from a file22:46
mordredyolanda: that said- I'm very excited that you and others are interested and willing to help us work on resolving it22:46
lifelessfungi: not just wheel22:46
*** Swami has quit IRC22:46
clarkbmordred: see comment on my test there shows that the str() may not do what you want it to do22:47
mordredclarkb: really?22:47
* mordred glares at SpamapS22:47
clarkbmordred: ya it will print None22:47
yolandamordred, yes, i was not talking about cores really, but do anything that we can to improve that communication and improve speed of reviews22:47
*** marun has quit IRC22:47
clarkbmordred: which you are explicitl trying to not do22:47
fungithinrichs: yeah, permissions look right. i'll double-check that there's not a second account for you or something22:47
clarkbmordred: but double check my manual test case22:47
yolandafrom our team we really feel we need to improve on that22:47
thinrichsfungi: I changed my email at some point.  Could that be the problem?22:47
mordredclarkb: no - you're right - I had a workaround for that in a previous patch, and I think I baby-with-the-bathwatered something22:48
clarkbmordred: np22:48
*** hyakuhei has quit IRC22:48
openstackgerritMerged openstack-infra/project-config: Run Gnocchi devstack tests for stable/1.0 against Kilo
*** hamzy has quit IRC22:49
*** hyakuhei has joined #openstack-infra22:49
*** ajo has quit IRC22:50
*** hyakuhei has quit IRC22:50
openstackgerritMonty Taylor proposed openstack-infra/shade: Call super in OpenStackCloudException
mordredclarkb: how about that ^^ ?22:50
lifelessclarkb: the '' is probably from grep -v '^#' $REPODIR/requirements/global-requirements.txt | while read req22:50
*** woodster_ has quit IRC22:50
mtreinishmordred: is that raised if OpenStack dies?22:50
lifelessmordred: why do you call pip wheel in a loop22:50
lifelessmordred: rather than 'pip wheel -r global-requirements.txt' ?22:50
*** xyang1 has quit IRC22:50
clarkbmordred: that should fix it +222:51
fungithinrichs: oh! you didn't sign the tag22:51
mordredclarkb: thanks!22:51
mordredlifeless: uhm ... there was a reason22:51
clarkblifeless: mordred yes the reason is if one fails you want to continue and do the others22:52
mordredlifeless: in some contexts it was to get as many things as possible - beacuse pip fails on the first22:52
thinrichsfungi: I was about to ask that!  So once I sign it, I shouldn't have any issues?22:52
*** dhritishikhar has joined #openstack-infra22:52
lifelessclarkb: pip itself does that22:52
clarkblifeless: it didn't before22:52
mordredlifeless: if it does, it's new that it does22:52
clarkblifeless: it would do everthing up to the fail then bail out iirc22:52
mordredit very much did that22:52
SpamapSmordred: +1'd as well.. And that is, btw, the answer to "why aren't you just passing extra_data to super" ;)22:52
fungithinrichs: you'll need to `git tag -d 2015.1.0rc1` to delete the current one in your local repo, and then try again using the command we have documented there in the infra manual22:52
mordredSpamapS: :)22:53
fungithinrichs: but yeah, otherwise everything looks okay with your account and permissions22:53
* mordred hands SpamapS a super22:53
lifelessmordred: clarkb: possibly a bug; the intent in pip is clearly to run them all and then fail at the end22:53
*** sdake_ has joined #openstack-infra22:53
mordredlifeless: neat22:53
*** achanda has quit IRC22:53
lifelessmordred:         # Return True if all builds were successful22:53
mordredI love bugs22:53
lifeless        return len(build_failure) == 022:53
lifelessand                 'Failed to build %s',22:53
lifeless                ' '.join([ for req in build_failure]),22:53
lifelessanyhow, that can wait22:54
thinrichsfungi: Thanks!  Didn't occur to me that signing would be required.  Will remember next time.22:54
lifelessit will fix that '' issue22:54
fungithinrichs: yep, if you look at your gerrit acl, the control is specific to signed tags only. unsigned annotated and lightweight tags are not allowed by that acl22:55
*** sdake__ has joined #openstack-infra22:55
fungithinrichs: that's mainly because people releasing software via git tags should be signing the tags, so having the acl that way helps keep us from accidentally forgetting to sign them22:56
*** sdake has quit IRC22:56
thinrichsfungi: makes sense.  Thanks again.22:56
fungithinrichs: any time!22:57
*** apmelton is now known as apmelton_away22:58
*** sdake_ has quit IRC22:59
fungiclarkb: regarding your comment on 176907, i was going to ask the same thing until i reread the commit message and realized that it seems like they're saying they expect to have multiple branches of gnocci they want to test against the same branches of openstack22:59
clarkbmordred: for the noqa at would it be beter to blanket disable that rule or are exceptions a special case?22:59
clarkbmordred: I am trying to remember if that is a hacking rule and I think it is22:59
clarkbwhich makes me think turning it off is ok22:59
mordredclarkb: I generally like that rule ... but also I didn't really want to shove exc. everywhere22:59
lifelessfungi: 2015-04-24 22:09:05.643 | + -upstream = -upstream23:00
lifeless2015-04-24 22:09:05.644 | /tmp/ line 8: -upstream: command not found23:00
lifelessfungi: that appears to be the problem23:00
fungiclarkb: so e.g., stable/1.0 and stable/2.0 against openstack stable/kilo but stable/3.0 against master or something23:00
fungilifeless: oh, it stripped the quotes!23:00
clarkbfungi: gotcha23:00
mordredclarkb: how about I follow up with a patch that either removes the rule or uses exc. everywhere?23:00
lifelessfungi: would -          if "{variant}" = "-upstream"; then23:00
lifeless+          if [ "{variant}" = "-upstream" ]; then23:00
lifeless help ?23:00
fungilifeless: er, nevermind. i forgot the word "test" there didn't i?23:00
lifelessfungi: or should I push the eval down into ?23:01
fungilifeless: or square brackets, yeah23:01
fungilifeless: well, you'll still need to get that parameter into the environment i think23:01
mordredclarkb: I need to rebase through that patch anyway23:01
greghaynesclarkb: youd like this
mordredclarkb: I'll just fix it this pass23:01
greghayneswrong link23:01
clarkbmordred: I just -1'd it for the rebase too23:01
openstackgerritlifeless proposed openstack-infra/project-config: Better shell.
fungilifeless: so if you want to do the eval in the script, you need to export a shell variable with that, which probably makes the script not stand alone as well23:02
greghaynesclarkb: that link23:02
lifelessfungi: the script barely is :)23:02
clarkbgreghaynes: \o/23:02
fungifair enough23:02
*** sarob has quit IRC23:02
lifelessnow I just need two cores to hammer that through23:03
clarkblifeless: looking23:03
lifelessand then go away for 45m and come back23:03
clarkbsilly test23:03
*** woodster_ has joined #openstack-infra23:03
fungilifeless: also my apologies for inflicting my terrible mailercode on you23:04
lifelessfungi: its ok, I should have applied my brain cap more too23:04
openstackgerritMonty Taylor proposed openstack-infra/shade: Split exceptions into their own file
openstackgerritMonty Taylor proposed openstack-infra/shade: Raise a shade exception on broken volumes
mordredclarkb: there ya go23:07
*** hemna is now known as hemnafk23:09
clarkbmordred: looking. also you hinted at infra cloud on list, where are we on getting hardware straightend out?23:09
clarkbI think it would be a neat exercise to just use the d-g deploment while we do a real thing23:09
mordredclarkb: yeah - your idea came up over lunch today23:10
mordredclarkb: although I think we've got to get the network configured at some point ...23:10
fungii might, you know, know someone who can configure network hardware, if that's the hold up23:11
mordrednah - that we've got23:11
clarkbright hardware and networking have to happen first, just wondering if that is moving at all23:11
mordredwe've been holding off on that because infra-cloud is not on the priority effort list yet23:12
* anteaya listens as she is also curious23:12
lifelessmordred: new hardware?23:12
mordrednope. same old hardware23:12
fungias in, is there out-of-band console access to the network gear and a diagram of how it's wired? the rest i'd be glad to help with23:12
mordredfungi: the missing part right now is an agreed upon network design from us23:12
*** ddieterly has quit IRC23:12
fungioh. got it23:13
fungiyes, that we will need23:13
mordredso maybe we'll come out of the summit with that23:13
mordredbecause once that's reconfigured, the ability for us to make interesting decisions and try things increases23:13
anteayahas it made it on the infra etherpad yet?23:13
fungimany snapchats of random whiteboard scribble23:14
clarkb/22 on a single l2 across all the nodes for VMs then like a /24 for the flat network?23:14
mordredclarkb: why two networks?23:14
clarkbhopefully we don't care much about it other than that? I suppose there may be lacp and other fun thigns to get into23:14
clarkbmordred: because nova net and neutron both expect that23:14
anteayanova-net does too?23:14
clarkbanteaya: yes23:14
* anteaya is trying to understand all the networking things23:15
mordredclarkb: I'm assuming you mean for control plane communication?23:15
fungiclarkb: i was hoping for a /4823:15
mtreinishclarkb: with a /24 be enough?23:15
fungifat chance, i know23:15
clarkbanteaya: mordred: well nova network calls it the "flat" network23:15
clarkbfungi: I think one of the two regions can ipv6 but not both23:15
mtreinishI have no idea how big this will be23:15
mordredclarkb: both of the regions can ipv623:16
clarkbmordred: oh thats good news23:16
mordredclarkb: but we need to file the paperwork to get a range23:16
fungimtreinish: /22 for the instances though, so that's "can haz 1k nodes" (i expect we'll be bottlenecked well below that for other resources anyway)23:16
clarkbbut mostly my networking with openstack has been related to what we do in the gate which is pretty vanilla for both nova net and neutron23:16
mordredclarkb: assume in general that I'm not up on the nova-network/neutron terminology - you described two networks - can you expand on what they are in normal people words?23:16
mtreinishfungi: ah ok, I read that backwards23:17
clarkbmordred: yes, one is a public network where all your public addrs go23:17
anteayaclarkb: your notes have been helpful for me, thank you23:17
mordredclarkb: and the other is a second network that is used for?23:17
fungimtreinish: and with a /24 for the metal, i seriously hope it's not so complicated we need more that that23:17
clarkbmordred: the other is a private side for VMs to talk to each other23:17
mordredclarkb: VMs or services?23:18
clarkbmordred: and now that I think about it mtreinish is right. for nova net you would need a /22 but neutron overlays so you wouldn't need any addrs assigned there23:18
clarkbmordred: VMs23:18
clarkbmordred: so think of it like how rax gives us two interfaces23:18
mordredclarkb: we don't want/need that23:18
mordrednot needed23:18
greghaynesclarkb: you sure its not services also? I think ec2 api goes over internal23:18
clarkbmordred: sure I have no clue how to make it not happen though23:18
thinrichsfungi: sorry to bother again.  Got that tag pushed, but now can't seem to push the branch.23:18
openstackgerritMerged openstack-infra/project-config: Better shell.
mordredI have it both on good authority from markmcclain that we don't need 2 - and also vexxhost has one and only one network23:19
clarkbmordred: like we have a gazillion config options and the ones I understand end up with a rax like setup23:19
thinrichsfungi: followed all the instructions at the docs you pointed me to.23:19
anteayaclarkb: I think that is part of what I am trying to understand as well23:19
mordredclarkb: yah - apparently the private is not needed at all23:19
greghaynesmordred: I keep hearing you dont need two, its not very well documented how to do it without two23:19
clarkbmordred: I am sure we can configure it but I think it was kevin benton who said "do it this way except its terribly documented sorry"23:19
mordredgreghaynes: sure - so maybe we can change that23:19
thinrichsfungi: trying to push directly from gerrit gave me the usual [remote rejected]23:19
clarkbgreghaynes: I don't expect to run metadata service23:19
mordredo hell no23:19
greghaynesclarkb: awesome23:19
thinrichsfungi: from I couldn't seem to find anyway to create a new branch—but I could see the list of branches.23:19
fungithinrichs: the branch creation works differently if you're not on the openstack release management team. you'll need have create permission on refs/heads/* in the acl and then use the web interface23:20
mordredI think we should design the thing we want - and then beat on PTLs to tell us how to configure it23:20
greghaynesok, well then doing one isnt a problem I think, that was my only hiccup in my recent experiment on only using one interface23:20
anteayamordred: oh perhaps ask them nicely?23:20
fungithinrichs: hich repo again?23:20
clarkbmordred: thats fair, it will also be slower which is why the d-g option sprang to mind23:20
mordredanteaya: :)23:20
fungier, whiwh23:20
clarkbmordred: we can have the bad thing that they want us to have23:20
thinrichsfungi: openstack/congress23:20
fungimy fingers hate me tonight23:20
clarkbmordred: then point at X,Y,Z as problems and say "this is what we actually want"23:20
mordredclarkb: sure - but I'd like to make sure that whatever network reconfig we ask the HP NOC to do supports the thign we want23:20
anteayafungi: ha ha ha23:20
mordredif there is also an extra 10. there, meh23:20
mordreddon't care23:21
clarkbmordred: yup, I would go with two networks23:21
clarkbmordred: it will be most flexible23:21
clarkband if we don't use one oh well23:21
anteayaclarkb mordred understanding why the make us have the bad thing would help me a lot23:21
anteayaclarkb mordred then the conversation about how to get the thing we want has a factual basis23:21
mordredanteaya: there is a "cloud-native" model that people like to talk about being the "good" way to do things23:21
fungithinrichs: yes, in the refs/heads/* section of you need a line like "create = group congress-release" to be able to do that23:22
mordredanteaya: it involves giving VMs private ip addreses by default and then getting a public IP to them by use of floating ips23:22
anteayamordred: who are these people and what documentation have they left in their wake?23:22
mordredanteaya: it's the "default" way things work23:22
anteayamordred: except in openstack apparently23:22
greghaynesAnother thing I just remembered - do we actually have two interfaces on all the hosts? I remember mention that some of the nics might have half their ports filled with infiniband instead of ethernet23:22
mordredanteaya: however - it breaks many things and is not advantageous to people actually using clouds23:22
mordredanteaya: nonononon23:22
mordredanteaya: sorry23:22
mordredanteaya: I radically misspoke23:22
* anteaya listens again23:23
mordredanteaya: there is an idea IN openstack that this is the right way tot od things23:23
mordredanteaya: we, in infra, disagree with it vehemently23:23
anteayawho is saying this in openstack?23:23
mordredbecause it makes things worse not better23:23
mordredanteaya: it's the default23:23
mordredanteaya: it's the way it works out of the box23:23
anteayawhose default?23:23
mordredanteaya: everythings23:23
clarkbanteaya: the documentations and devstack23:23
lifelessas much as openstack can be said to work out of the box23:23
lifelessat all23:23
mordredlifeless: well, yeah23:23
anteayanova-net's default or neutron's default?23:23
clarkbanteaya: both23:23
* anteaya listens again23:24
clarkbanteaya: this is how we test both neutron and nova net in the gate23:24
mordredit's a bonghits terrible design23:24
anteayaah okay thank you23:24
clarkbanteaya: and we don't toggle any devstack bits to set it up that way23:24
mordredit's not a design to make somethign god23:24
anteayaclarkb: thank you I understand that part now23:24
mordredit's a design to conserve IPv4 addresses ina  public cloud23:24
anteayawhich with ipv6 we don't care about, yeah?23:24
mordredit makes applications harder to work with23:24
mordredwell, we don't care about ipv4 conservation23:24
mordredwe have plenty23:24
anteayaso what do we (infra) want?23:25
fungialso, any day now, ipv6. then we can just stop using ipv4 for things like job workers if we want23:25
mordredwe want vms to get an ip directly allocated to them on the network23:25
anteayafungi: yay23:25
mordredwe don't want floating ips23:25
anteayamordred: which from my reading is what dhcp does, yes?23:25
mordredthey are a nightmare23:25
mordredanteaya: yes23:25
clarkbfungi: I am still keeping an eye on the turn on ipv6 by default changes. need one fix in nova first23:25
clarkbfungi: but that should happen early this cycle (liberty)23:25
*** sigmavirus24 is now known as sigmavirus24_awa23:25
mordredit's this great technology that was invented back in the 90s to dynamically configure hosts :)23:25
fungifor public clouds it'll likely be hard to convince them to give us instances with no global ipv4 but working ipv6. if we run our own though, possibility23:25
clarkbmordred: 80s even23:26
mordredclarkb: ++23:26
mordredfungi: btw - vexxhost just gives a public ipv4 and no private network at all- I believe runabove is the same23:26
anteayaso does either networking solution give us what dhcp does?23:26
clarkbmordred: yes my runabove instance was that way23:26
anteayaor is close?23:26
fungimordred: sure, but point being we can potentially also accomodate v6-only providers23:27
mordredanteaya: well, we can use dhcp with either neutron or nova-net23:27
mordredfungi: ++23:27
mordredanteaya: we will definitely run dhcp23:27
fungimordred: and if wherever we're colocating the gear tells us we can't have more than a /24 of ipv4 for example, we could still spin up our own v6-only instances23:27
openstackgerritMerged openstack-infra/devstack-gate: Make multinode first class with ansible
clarkbfunny story about the bug preventing ipv6 in devstack by default23:28
fungiand tell nodepool "your workers are over there, on the new internet, full of happy ipv6 elves and no ipv4 trolls"23:28
mordredclarkb: ^^ woot!23:28
clarkbit breaks floating IP attachment23:28
clarkbbecause it tries to nat ipv4 to ipv623:28
clarkb(more evidence that maybe this was a bad design)23:28
fungiyou just said nat. not map?23:28
clarkbfungi: NAT is how floating IPs work23:28
anteayaso if we can run dhcp with both nova-net and neutron how am I not seeing that this is a solution that anyone can use?23:29
mordredit's amongst the many reasons they're bad23:29
fungioh, got it. missed you said floating ip23:29
anteayawhy all the hatred from users if they can do this themselves?23:29
mordredI mean, who in the world wants to run a real service on an internet connection that looks like their home DSL connection23:29
* anteaya is still trying untangle and understand23:29
clarkbanteaya: the default setups use dhcp, thats not the issue23:29
anteayaclarkb: oh okay23:29
anteayawhat is the issue?23:29
clarkbanteaya: the issue is they force you to have two sets of networking and use NAT to map between them23:29
clarkbanteaya: this results in extra api calls as a user because you have to do that manuall23:30
anteayacan you configure that so you don't have to have that?23:30
anteayaor not configurable?23:30
clarkbanteaya: so your code is more complicated and error prone (we leak floating IPs)23:30
openstackgerritTim Hinrichs proposed openstack-infra/project-config: ACL changes for Congress create-reference
mordredanteaya: only when we run our own cloud23:30
lifelessmordred: well23:30
mordredanteaya: we cannot configure public clouds we consume to do this23:30
lifelessmordred: as a counterpoint to your rage23:30
anteayamordred: oh23:30
clarkbanteaya: mordred however, the other options are basically undocumented23:30
mordredanteaya: which is why we have rage23:30
mordredclarkb: sure23:30
anteayaclarkb: ah ha23:30
mordredclarkb: I'll write docs23:30
clarkbanteaya: mordred so while you can strike it out on your own you are really on your own23:30
lifelessmordred: the largest most commercially successful clouds work with a private+public split23:30
anteayamordred: yes I can feel the rage23:31
mordredlifeless: sure23:31
*** dhritishikhar has quit IRC23:31
mordredlifeless: for the reason that I stated above23:31
clarkblifeless: that doesn't make it user friendly23:31
mordredit doesn't make it a good design23:31
anteayaclarkb: do you think better documentation might actually be part of the solution to finding less rage here?23:31
mordredit makes it the crappy design that people are forced to deal with because that's what bezos gives them23:31
clarkbanteaya: yes, I think it will help quite a bit so that deployers know that they can do things in different and potentially better ways for them23:31
mordredand so, stockholm syndrome kicks in, and people assume it's "right"23:31
openstackgerritTim Hinrichs proposed openstack-infra/project-config: ACL changes for Congress create-reference
lifelessmordred: I'm not arguing right23:32
mordredit's never been a feature a user has _wanted_23:32
anteayaclarkb: okay thank you, this gives me some hope23:32
mordredit's a feature that can be dealt with23:32
lifelessmordred: but its clearly not such a burden that it makes the product undesirable23:32
anteayamordred: yes stockholm syndrome, sigh23:32
mordredto maximize someone else's profits23:32
mordredlifeless: I didn't say it was23:32
mordredlifeless: I said that, in this case, I have a choice23:32
mordredbecause openstack gives me that freedom23:32
mordredso I'm going to use it23:32
mordredand I'm going to show that you do not have to be bound by what amazon deigns to give you23:32
mordredin fact23:33
*** jerryz has joined #openstack-infra23:33
openstackgerritMatthew Treinish proposed openstack-infra/subunit2sql: Switch functools partials usage to kwargs
mordredit might be the most important feature of openstack23:33
mordredthat it gives people the freedom to not be bound by the design choices that were made to maximize the profits of a bookseller23:33
mordredand instead make design choices taht are right for their workload(s)23:33
mordredthat _might_ be a floating ip model23:33
mordredwhich is neat if it is23:33
openstackgerritTim Hinrichs proposed openstack-infra/project-config: ACL changes for Congress create-reference
clarkbalso woot for better d-g multinode support23:34
clarkbpleia2: ^ thanks a bunch23:34
mordredbut assuming that floating is the only or best way to go just beause AMZN did it that way is bonghits23:34
mordredclarkb: so excited!23:34
*** akshai has joined #openstack-infra23:34
anteayaclarkb pleia2 yay23:34
anteayamordred: is that the reason floating ips are a thing?23:35
anteayabecause amazon does it that way?23:35
*** dannywilson has quit IRC23:35
fungiwell, in some cases also because of the same reasons that amazon did it23:35
mordredanteaya: yes23:35
anteayafungi: ip conservation23:36
*** ChuckC has quit IRC23:36
fungie.g. to conserve global ipv4 assignments on the assumption that you may boot 20 instances in your tenant only 3 of which run webservers23:36
anteayawhich we don't care about any more, if we ever did23:36
fungiand for most of the world, if it's not a webserver it's not part of "teh internets"23:36
clarkbmordred: your exception move failed some unittests so its not perfect yet23:36
*** shashankhegde has quit IRC23:36
anteayafungi: yes, I'm hearing that23:36
mordredclarkb: I want to cry23:36
clarkbmordred: sorry, I did grep through and that looked fine23:37
*** shashankhegde has joined #openstack-infra23:37
clarkbmordred: so I Think it must be in the tests themselves23:37
anteayathank you for your patience in helping me to understand the issue23:37
anteayanot that I entirely do but this helps add23:37
mordredclarkb: I thought I'd run them locally ...23:37
* mordred re-runs23:37
clarkbgoing back to what infra cloud will need, it will need at least one network with publicly routable addresses (ipv6 only should be ok)23:37
mordredclarkb: well, it's actually easier/less work at first to just get a /2223:38
jrollI'm missing context, but floating IPs are also a thing for HA and whatnot, no? (maybe this isn't inherently just the issue)23:38
mordredgetting the IPv6 range will be more work, amusingly enough23:38
mordredjroll: they really aren't23:38
fungiyeah, i totally expect this one to start with v4 (or with our luck be v4 only for a while)23:38
clarkbjroll: you should not rely on floating Ips for HA23:38
clarkbjroll: your NAT tables are not replicated or anything, its basically a terrible ha model23:38
mordredjroll: they're not nearly good enough for that - if you want IP-based HA, you need to do a load-balancer or something23:38
*** dhritishikhar has joined #openstack-infra23:38
clarkbmordred: exactly23:38
jrollmordred: clarkb: I'm thinking like HA load balancers23:38
fungijroll: ha needs actual load balancers with failure detection23:39
mordredjroll: they're not good enough for that23:39
clarkbjroll: floating IPs won't help there23:39
mordredthere is one use case they are good for23:39
jrollthen how do you do it?23:39
* jroll is just being curious here23:39
* anteaya is also glad jroll is curious23:39
mordredthat is having a floating ip with DNS attached to it that yuo can use to minimize downtime across a server migration ofa  pet-like service23:39
lifelessjroll: so there are three related things I think23:39
mordredwhich is the funny thing23:39
*** pc_m has quit IRC23:39
mordredthe one main use case in which the implementation actually is useful23:39
clarkbjroll: anycast? I don't think openstack has a good answer there23:39
lifelessjroll: one is rapidly moving an IP from one host to another, used by *some* HA patterns23:39
clarkbjroll: but floating IPs are a bad answer23:40
mordredis the LEAST "cloud" design23:40
fungijroll: in the openstack case, with a separate api service to instantiate a lb cluster of some sort23:40
lifelessjroll: another is making some hosts unreachable from the i/n23:40
lifelessjroll: a third is making hosts aware of all the IPs that they have23:40
fungi(also done at least as well with packet filters)23:40
lifelessNAT based floating ips give you the first two but not the third23:40
*** shashankhegde has quit IRC23:41
lifelessexcept via extras like the metadata service or config-drive23:41
jrollright, so for example I run haproxy in front of an API service - I use corosync and a vip/floating IP to do failover if one crashes and burns23:41
clarkblifeless: and only metadata service will update23:41
jrolland I guess I'm wondering if/why that doesn't work23:41
lifelessthe former is dynamic, except our impl is unperformant, and constrained to be little better than config drive23:41
mordredjroll: but a floating IP is not a vip23:41
lifelessand config drive is static, but folk prefer it because it has less security ramifications23:41
clarkblifeless: and its reliable23:42
mordredyes - it works much better23:42
jrollok, that's where I was mistaken23:42
mordredof course, DHCP also informs a host of its ip ... :)23:42
*** akshai has joined #openstack-infra23:42
jrollvip vs floating ip (I sometimes hear of vips called floaters)23:42
lifelessso, vip's require appropriate acls in the network layer23:42
mordredjroll: yah - it's easy to think they're the same23:42
lifelessbutthats doable23:42
fungii have hope that _some_ day i can provision a network in openstack that will let me actually route additional ip addresses to use for things like real clustering/takeover/sharing protocols23:42
jrollok, thanks for helping me get that :)23:43
lifelessthere are other ways to deliver 1)23:43
fungiand to be able to control ip addresses by actually configuring them within the context of my virtual machines23:43
clarkbfungi: I think it would just be a matter of teaching neutron routers that some IPs are special and will be managed outside the neutron IPAM service23:43
mordredfungi: right?23:43
lifelesssuch as - for services that need it - running a vip23:43
clarkbfungi: a "leave this IP the hell a lone" api call23:43
lifelessor ip unicast23:43
lifelessexport your hosts via BGP as service records23:44
clarkbthen you could do VIPs within your service23:44
lifelessor a cloud provisions haproxy23:44
lifelessor <...>23:44
lifelessthe second - filters please.23:44
fungihowever, i'm probably a luddite who simply hasn't embraced the "cattle not pets" design philosophy23:44
mordredfungi: right - EXCEPT23:44
lifelessthe third, having a vif per IP and dhcp on each vif, can get you it with no magic in the VM23:44
mordredfungi: floating IPs are only useful for pets23:44
lifelessno awareness of which cloud23:44
mordredfungi: they have no value at all for cattle23:45
lifelessmordred: why do you say that?23:45
anteayafungi: yes I definitely picture you as a luddite23:45
lifelessmordred: they are AFAICT on different dimensions23:45
jrolldevil's advocate: running mail servers on the cloud, you want to keep your IPs, floating IPs are useful23:45
mordredlifeless: beacuse the value is taht you can move the ip to another host - if you have cattle, that concept is not needed23:45
jrollmaybe mail servers are a pet problem :P23:45
mordredlifeless: if an ip is not special, there is no need to migrate it23:45
fungianteaya: sometimes i like to use "mendicant" instead23:45
anteayaha ha ha23:45
mordredjroll: nope - you can scale out clusters of those23:45
lifelessmordred: so 'ip is special' and 'host is special' are IMO very different23:45
lifelessmordred: if the ip is the ip of one of your cattle haproxy23:46
fungijroll: twitter is the new e-mail. see, you're a luddite too. embrace the cattle future23:46
jrollmordred: sure, but you need a clean external IP somehow, that may need to bounce somewhere else23:46
mordredjroll: not really23:46
jrollfungi: I bet twitter uses floating ips ;D23:46
clarkbtwitter uses mesos iirc23:46
mordredjroll: you can TOTALLY have multiple non-related incoming23:46
jrollyeah, I kid about twitter23:46
mordredjroll: you do not need one IP23:46
clarkbwhich probably has its own crazy VIP implenetnation on top of it for routing traffic arbitrarily23:47
jrollmordred: sure, but you don't want to constantly recycle IPs either23:47
lifelessits largely about the clients23:47
lifelessmost clients are not prepped for rapidly changing ip's on their endpoints23:47
mordredjroll: I don't think you need to constantly recycle - just because an IP isn't special doesn't mean it needs to change hourly23:47
lifelessnothing to do with how cattle-or-pet the endpoints are23:47
* fungi wants to constantly recycle ips23:47
mordredtotally agree with lifeless on that23:47
mordredit's not about changing them constnatly23:47
jrollmordred: recycle frequency depends how crappy your cloud is :P23:48
jrollbut yeah, I get your point23:48
mordredjroll: well, none of this is going to fix the cloud sucking :)23:48
mordredin fact - it turns out each of these things are complexity layers23:48
* jroll is joyful it's friday and he can stop dealing with clouds for a couple days23:48
mordredso if your cloud is crappy, NOT having floating ips means one less thing to break23:48
openstackgerritClint 'SpamapS' Byrum proposed openstack-infra/shade: Do not cache unstead state images
mordredlifeless: I do think there are some uses for them - but I think most of those are really things that should be more opt-in23:49
clarkb(if the gate testing is any evidence yourfloating IPs will break)23:49
mordredclarkb: ++23:49
lifelessmordred: I can get behind that23:49
clarkbprotip do not use nova shelve23:49
SpamapSclarkb: ^ my preferred answer to "shouldn't we invalidate outside the loop" :)23:49
clarkbSpamapS: looking23:49
lifelessmordred: I just don't think that an argument against them with significant conflation is a solid one23:49
lifelessbecause the people listening to the argument will be aware of this23:50
jrollclarkb: lol23:50
lifelessbetter to separate out the concerns23:50
SpamapSclarkb: actually let me re-state.. the answer to that question is _yes_, but we shouldn't have to invalidate _inside_ the loop.23:50
mordredlifeless: yup. I'm most appending "for me" or "for infra" on to most of my blanket assertions in here23:50
clarkbSpamapS: right looks like we are telling it to only cache once we have reached steady state23:50
jrollmordred: so is your actual grunt here that floating IPs are being forced onto you?23:50
clarkbSpamapS: which seems sane to me23:50
jrollI haven't read all the scrollback23:50
lifelessmordred: thats useful to know, but still can't copy-paste your argument to anyone else :)23:51
clarkbjroll: thats a major part of it at least for me23:51
mordredjroll: well, they have been for a while - HP uses them ... which means I have to deal with them23:51
fungijroll: basically23:51
clarkbjroll: we have no other option in hpcloud23:51
mordredjroll: and it's much harder for no benefit23:51
mordredso - for infra-cloud23:51
clarkbjroll: and we leak floating IPs and all kinds of nasty23:51
mordredwe do not intend to use them23:51
mordredbecase they do not benefit us23:51
jrollI haven't used HP cloud, so I wasn't aware you were required to use them23:51
anteayamordred: I take every statement you make to be appended with 'for me'23:51
mordredanteaya: :)23:51
SpamapSclarkb: right, because caching things in 'saving' state is just going to result in a very out of date cache.23:51
fungijroll: hpcloud gives you rfc-1918 "public" interfaces and then you make a separate api call to bind a globally routable fip23:51
clarkbSpamapS: yup23:51
jrollfungi: that's annoying23:51
fungijroll: yeah. yeah it is23:52
anteayajroll: the largest issue I see is that people are forced to use things they don't want to use23:52
clarkbjroll: but thats neutrons default setup23:52
jrollit's like legos23:52
anteayajroll: that makes them unhappy23:52
clarkbjroll: if you follow the docs and/or use devstack this is what you get23:52
SpamapSThat's neutron's default setup that happens to emulate EC2. ;)23:52
*** sdake__ has quit IRC23:52
openstackgerritMonty Taylor proposed openstack-infra/shade: Split exceptions into their own file
mordredclarkb: ^^ found it23:52
SpamapSOr does EC2 still automatically give you a routable IP? (I had heard that was going away)23:53
fungialso openstack conversations should have a godwin's law variant involving amazon ec223:53
*** thinrichs has quit IRC23:53
clarkbSpamapS: it doesn't matter we have the freedom to make something better23:53
SpamapSclarkb: ++23:53
*** achanda has joined #openstack-infra23:53
mordredSpamapS: did you see I split the exceptions - just for you?23:54
anteayaclarkb: so again having better docs looks like a potential way forward?23:54
mordredSpamapS: you have suggestions on what exception handling tests should look like?23:55
SpamapSmordred: I did, and I promise to nuzzle you for that.23:55
mordredSpamapS: I can add that to my list next23:55
anteayaso deployers of public clouds are more likely to deploy them in a way that is useful to us?23:55
mordredanteaya: actually, we already have public clouds deploying the way we want23:55
anteayawe do23:55
mordredanteaya: vexxhost and runabove both have good configs23:55
anteayarunabove sorry23:56
*** wenlock_ has quit IRC23:56
*** ChuckC has joined #openstack-infra23:56
anteayado we know what their configs are?23:56
anteayacan we get them?23:56
SpamapSmordred: mock_client.something.side_effect = Exception('whatever you want to emulate')23:56
mordredwe've got project PTLs - we can figure it out I'm sure23:56
mordredanteaya: I mainly want to start fighting the theory23:56
SpamapSmordred: and you probably want to assert log contents23:56
mordredthat "floating IPs are the 'right' way to do it"23:56
anteayamordred: well you enjoy that23:56
SpamapSOk, I believe weekend just started23:57
anteayaI'm interested in getting users something they are more likely to use23:57
mordredanteaya: and I want to make sure that we help get the documentation to show what you need to do to ge a cloud that behaves this way23:57
mordredanteaya: ya23:57
anteayamordred: me too23:57
* SpamapS also believes a slight fever is besetting him.. damnit23:57
mordredanteaya: or - I want to make sure people know they have the freedom23:57
anteayaSpamapS: :(23:57
mordredSpamapS: gross!23:57
mordredanteaya: so that they dont' just hear people telling them that they MUST rewrite all of their apps to use cloud23:57
anteayamordred: well helping them do it the way they want proves to them they have the freedom23:57
SpamapSYeah, or its just the massive grey cloud asserting low pressure and general murkiness on LA23:57
clarkbmordred: fwiw I think they would be much more palatable if the implementation of the floating ip was to assign an IP to a real interface and not have NAT running23:57
mordredbecaus that's just not true at all23:57
anteayamordred: you are welcome to lead the parade23:57
mordredclarkb: ++23:57
mordredclarkb: I agree23:57
mordredclarkb:  I would like a floating-ip option that behaved more like a traditional vip23:58
mordredthat would be neat23:58
*** sabeen3 has quit IRC23:58
clarkbbecause then the node is aware of its IP and can be smart and you don't have to work around the crazyness of NAT23:58
clarkbmordred: ya23:58
SpamapSclarkb: I believe that invites ARP storms from your users though.23:58
SpamapSclarkb: even if they won't work.. users will swear they do.23:58
clarkbSpamapS: not i you are controlling the assignment23:58
anteayamordred: from what I'm reading they become unhappy any time anyone tells them they must do anything23:58
clarkbSpamapS: and otherwise your users can arp storm all day23:58
anteayathey want to do what they want to do23:58
mordredok. I'm giong to go eat pork23:59
clarkbSpamapS: the same thing that prevents me from claiming I have an ip address today would be used in that case23:59
*** achanda has quit IRC23:59
clarkbmordred: I get to do that tomorrow23:59
*** shardy_ has joined #openstack-infra23:59
clarkbI bought >26lbs of meat yesterday23:59
SpamapSclarkb: yes, except that is now 100% always malicious. :)23:59
SpamapSclarkb: you are now inviting it to also be ignorant.23:59
anteayamordred: enjoy pork23:59
SpamapSJust stating the psychological ramifications, not any actual reason it will break something.23:59
clarkbSpamapS: rgr23:59

Generated by 2.14.0 by Marius Gedminas - find it at!