Monday, 2015-04-13

*** sdake has joined #tripleo00:30
openstackgerritMonty Taylor proposed openstack/diskimage-builder: Add element to process config-drive network info
mordredSpamapS, greghaynes: ^^00:31
*** sdake_ has quit IRC00:34
*** eghobo has joined #tripleo00:34
*** sdake_ has joined #tripleo00:46
*** sdake has quit IRC00:49
*** sdake has joined #tripleo00:52
*** sdake_ has quit IRC00:56
*** eghobo has quit IRC00:58
*** eghobo has joined #tripleo00:59
*** eghobo has quit IRC01:08
*** yamahata has joined #tripleo01:17
greghaynesmordred: whered you split the simple-init out to?01:17
mordredgreghaynes: also, patch is up to suck it in to gerrit01:18
greghaynesah, awesome01:19
*** eghobo has joined #tripleo01:23
clarkbalso note gitorious is almost01:44
clarkber almost dead01:44
greghaynesyea, ill write up some dib magic to install from gitorious for now though01:44
greghaynesshould rework tests too01:45
*** eghobo has quit IRC01:49
*** Marga_ has quit IRC01:50
*** weshay has quit IRC02:03
*** sdake_ has joined #tripleo02:11
*** sdake has quit IRC02:14
*** yamahata has quit IRC02:20
*** Marga_ has joined #tripleo03:11
*** Marga_ has quit IRC03:12
*** subscope_ has joined #tripleo03:12
*** Marga_ has joined #tripleo03:12
*** sdake_ has quit IRC03:25
*** subscope_ has quit IRC04:43
*** tzumainn has quit IRC04:45
*** maufart has joined #tripleo04:45
*** pradeep has joined #tripleo04:45
greghaynesHTF do you do code review in gitorious04:47
greghaynesok, figured it out. I hope that is one of the last times I have to experience that04:52
greghaynesand looks like the tests 'just work'? \O/04:53
*** ishant has joined #tripleo04:53
StevenKgreghaynes: gitorious is that delightful to use? :-P04:54
*** eghobo has joined #tripleo04:54
greghaynesI cant imagine why it never caught on as much as some competitors04:54
greghaynesmordred: so thought - maybe we should rename the element glean also05:08
*** maufart has quit IRC05:09
*** masco has joined #tripleo05:18
*** jtomasek has joined #tripleo05:32
*** yuanying has quit IRC06:04
openstackgerritgreghaynes proposed openstack/diskimage-builder: Add element to process config-drive network info
greghaynesmordred: ^ also added some logic so we can install from git (and therefore test)06:06
*** yuanying has joined #tripleo06:08
*** dsneddon has quit IRC06:11
*** adrianopetrich has joined #tripleo06:20
*** jcoufal has joined #tripleo06:26
*** eghobo has quit IRC06:30
*** rwsu has joined #tripleo06:34
*** eghobo has joined #tripleo06:40
*** yamahata has joined #tripleo06:42
*** jprovazn has joined #tripleo06:45
*** ukalifon1 has joined #tripleo06:46
*** akrivoka has quit IRC06:46
*** eghobo has quit IRC06:51
*** yuanying has quit IRC06:54
openstackgerritDaneyon Hansen proposed stackforge/kolla: Adds Neutron Agents Support
*** yuanying has joined #tripleo07:21
*** jistr has joined #tripleo07:24
*** yamahata has quit IRC07:24
openstackgerritYanis Guenane proposed stackforge/puppet-tripleo: Enable access to HAProxy stats page
*** chlong has quit IRC07:25
*** yuanying has quit IRC07:28
*** yuanying has joined #tripleo07:31
*** Marga_ has quit IRC07:32
*** yuanying has quit IRC07:33
openstackgerritDaneyon Hansen proposed stackforge/kolla: Adds Neutron Agents Support
*** gfidente has joined #tripleo07:37
*** Goneri has joined #tripleo07:37
*** ifarkas has joined #tripleo07:41
*** yuanying has joined #tripleo07:48
*** yuanying has quit IRC07:51
*** yuanying has joined #tripleo07:52
*** yuanying has quit IRC07:54
*** yuanying has joined #tripleo07:59
*** yuanying has quit IRC08:02
*** mmagr has joined #tripleo08:05
*** hewbrocca has joined #tripleo08:06
*** yuanying has joined #tripleo08:07
tchaypoisn’t gitorious about to go bye-bye?08:09
*** lucasagomes has joined #tripleo08:09
tchaypo - " will stay online until the end of May 2015 to give people time to migrate their repositories."08:09
tchayposo my answer is “don’t bother doing code review on gitorious, it’s going down the memory hole in a few days anyway"08:10
tchaypothis answer may not work for or appeal to some use cases.08:10
*** dtantsur|afk is now known as dtantsur08:14
*** pblaho has joined #tripleo08:14
*** lsmola_ has joined #tripleo08:16
*** stendulker has joined #tripleo08:16
*** bvandenh has joined #tripleo08:24
*** derekh has joined #tripleo08:32
derekhNg: just saw those ssl cert emails, are you renewing that cert ?08:36
*** pradeep has quit IRC08:37
openstackgerritYanis Guenane proposed openstack/tripleo-heat-templates: galera: Allow installer to setup galera cluster
openstackgerritYanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration
openstackgerritYanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration
openstackgerritYanis Guenane proposed openstack/tripleo-heat-templates: Enable use of coordination_url in ceilometer
*** pelix has joined #tripleo08:53
*** athomas has joined #tripleo09:01
Ngderekh: yeah I'll grab a new one, then make all of this be slagle's problem :D09:16
derekhNg: cool, thanks, once you have it, can you put it on the rh1 server somewhere (I guess you still have access) and I'll take it from there09:17
derekhNg: If you have any problem accessing the server just shout and I'll take a look09:19
spredzyjistr, when you have time if you could re-take a look at I addressed the issue I -1'ed it for09:20
jistrspredzy: sure09:20
spredzyjistr, thanks09:21
*** mmagr has quit IRC09:27
Ngderekh: ok, new there's a new CRT and chain CRT in ~ng/ on the RH bastion09:39
Ngderekh: (the .key stays the same, the new crt/chain files have -2015 in the filename)09:39
*** mmagr has joined #tripleo09:41
derekhNg: cool, thanks09:41
*** pradeep has joined #tripleo09:42
openstackgerritShivanand Tendulker proposed openstack/diskimage-builder: Add element ubuntu-signed to provide signed kernel
openstackgerritShivanand Tendulker proposed openstack/diskimage-builder: Fix to load only signed kernel in UEFI secure boot
derekhNg: install and working10:03
Ngderekh: woot \o/10:03
derekhCI jobs have restarted10:03
*** sdake has joined #tripleo10:09
*** athomas has quit IRC10:11
*** amitpp has joined #tripleo10:14
*** athomas has joined #tripleo10:18
*** amitpp has quit IRC10:22
*** sdake_ has joined #tripleo10:28
*** akrivoka has joined #tripleo10:29
*** sdake has quit IRC10:31
*** stendulker has quit IRC10:50
*** sdake_ has quit IRC10:53
*** yuanying has quit IRC10:58
*** ishant has quit IRC11:01
*** masco has quit IRC11:05
*** rhallisey has joined #tripleo11:12
jistrspredzy: just a nit there, the patch is missing a dependency on a patch which adds galera_bootstrapped fact, otherwise lgtm. Maybe we could also add a comment above the exec which does daemon-reload that we need it because we're altering mariadb systemd unit. I think someone might be surprised why we need it when reading the code top-down...11:29
jistri'll check out the two patches and try to deploy 3 controllers11:32
*** ukalifon3 has joined #tripleo11:34
*** ukalifon3 has quit IRC11:34
*** ukalifon2 has joined #tripleo11:35
*** ukalifon5 has joined #tripleo11:37
*** ukalifon1 has quit IRC11:38
*** ukalifon2 has quit IRC11:40
openstackgerritYanis Guenane proposed openstack/tripleo-heat-templates: galera: Allow installer to setup galera cluster
spredzyjistr, thanks just fixed it ^11:41
openstackgerritYanis Guenane proposed openstack/tripleo-heat-templates: galera: Allow installer to setup galera cluster
*** lucasagomes is now known as lucas-hungry11:47
*** weshay has joined #tripleo11:47
*** spredzy is now known as spredzy|afk11:47
*** jdob has joined #tripleo11:56
*** sdake has joined #tripleo12:00
*** sdake_ has joined #tripleo12:01
*** ukalifon5 has quit IRC12:03
*** sdake has quit IRC12:04
*** ukalifon1 has joined #tripleo12:09
*** pradeep has quit IRC12:10
*** tzumainn has joined #tripleo12:12
*** EmilienM|afk is now known as EmilienM12:19
*** morazi has joined #tripleo12:21
*** dtantsur is now known as dtantsur|brb12:25
*** sdake_ has quit IRC12:25
*** sdake has joined #tripleo12:25
mordredgreghaynes: wow. gitorious really sucks doesn't it?12:26
*** dprince has joined #tripleo12:26
*** sdake_ has joined #tripleo12:27
*** spredzy|afk is now known as spredzy12:28
*** sdake has quit IRC12:31
spredzyjistr, any luck ?12:33
jistrspredzy: it failed, i'm looking into the cause12:34
spredzyjistr, arf, the cause might be RabbitMQ12:34
spredzythe broker didn't start properly, tables_not_present, linked to mnesia12:34
gfidentejistr, ^^ that's what we were talking about12:35
jistrindeed it's that12:36
gfidentejistr, did it fail on all nodes?12:36
spredzyjistr, this is what I apply for testing things and not being blocked by this atm
*** rlandy has joined #tripleo12:36
spredzynot nice, but at least RabbitMQ does not fail and I can test what I am deploying12:37
jistrgfidente: yup all nodes12:38
gfidentejistr, all three nodes failed to start it?12:39
jistri'll try to investigate a fix for this, if i can't come up with it, i'll use the patch to work around12:40
jistrbut we'll probably have to fix it eventually anyway12:41
jistrspredzy: thx12:41
*** lucas-hungry is now known as lucasagomes12:41
jistri'll report the bug12:43
gfidentejistr, sorry :(12:44
jistrgfidente: hehe nothing to be sorry for imo :)12:45
openstackgerritGiulio Fidente proposed openstack/tripleo-puppet-elements: Add pacemaker and pacemaker_remote packages and puppet module
*** pradeep has joined #tripleo12:50
*** jroll has quit IRC12:50
*** jroll has joined #tripleo12:50
*** openstackgerrit has quit IRC13:00
*** openstackgerrit has joined #tripleo13:03
spredzygfidente, jistr I am investigating this afternoon13:04
openstackLaunchpad bug 1443426 in tripleo "RabbitMQ cluster fails to start (tables_not_present)" [High,Triaged]13:04
jistrspredzy: me 213:04
spredzyjistr, ack. Let's keep each other posted if any progress13:04
*** trown|outttypeww is now known as trown13:04
*** Marga_ has joined #tripleo13:05
gfidentespredzy, jistr that's cool from me, I started with the basic pacemaker stuff, will add you on review13:06
EmilienMgfidente: I was about to look today13:06
EmilienMgfidente: are you going to push puppet code? (pacemaker bits)13:07
*** spzala has joined #tripleo13:07
gfidenteEmilienM, not until I need it, currently I just added it into the images and I'm bassing the basic params for the clustering of nodes13:07
gfidenteI was touching with jistr the issue of how to provide pacemaker with the details for the fencing agents13:08
EmilienMgfidente: cool. Ping me back when you got these bits, and I'll push puppet things13:08
gfidentethat seems an area where some puppet code will turn useful13:08
EmilienMmost of puppet code is already in upstream modules13:08
EmilienMshould be straightforward13:08
EmilienMspredzy: fyi ^13:08
EmilienMspredzy: I plan to use
gfidenteEmilienM, okay that's something we might want to check because I'm pulling in redhat-openstack/puppet-pacemaker for now instead13:10
openstackgerritJan Provaznik proposed openstack/tripleo-common: Add basic project structure
gfidenteEmilienM, I think they can work together13:11
spredzyEmilienM, ack13:11
gfidenteEmilienM, I will add that to the list as soon as I have a service to be configured in pcs13:12
jistrjprovazn: ^^ we'll need a patch for the new repo to openstack-infra config i guess (currently there are no cores set up, i can't +2)13:16
*** radez_g0n3 is now known as radez13:18
jprovaznjistr: looking13:19
openstackgerritShivanand Tendulker proposed openstack/diskimage-builder: Fix to load only signed kernel in UEFI secure boot
jistrjprovazn: ah it's set up for a separate core group tripleo-common-core instead of tripleo-core (not sure if that's desired or not)
jprovaznjistr: yes - my fault - I just followed howto when registering that lib w/o realizing that tripleo group should be set13:23
jistrspredzy: so we looked into the bug a bit earlier with gfidente. It seems that it's some kind of race condition on stopping/starting rabbitmq, triggered when creating a fresh erlang cookie for rabbitmq.13:25
gfidentejistr, I shared that with spredzy as well but I think he mentioned having seen it failing despite the sleep13:26
hewbroccawait, that bug is forever old13:26
hewbroccatalk to eck13:26
spredzygfidente, jistr yes the sleep didn't solve anything13:26
jistrhewbrocca: yeah i've googled that but it also looked like it's forever fixed13:26
hewbroccaI feel like he fixed it in the init script13:27
spredzyOne thing I was curious about (about to try) run rabbitmq on step 2 in the bootstrap node and on step 3 on the other nodes13:27
gfidentejistr, I added you to two pacemaker changes, they work for me locally :)13:27
hewbroccaAre you guys using the rabbit resource agent? I thought it just dealt with all this nonsense.13:27
gfidentehewbrocca, not yet indeed13:27
jistr bug 1059913 in rabbitmq-server "Race condition creating .erlang.cookie" [Unspecified,Closed: errata] - Assigned to rjones13:27
gfidente^^ pacemaker stuff going to be added13:27
* hewbrocca ceases manager randomization13:28
hewbroccasorry guys13:28
hewbroccacarry on13:28
spredzyjistr, gfidente what's anoying is the randomness13:28
spredzylike reight know it just worked.13:28
*** prad has joined #tripleo13:28
spredzyjistr, thanks for the link13:31
TheJulia /win 713:34
EmilienMgfidente: not really, we should/will rely on
EmilienMand not puppet-pacemaker13:34
* TheJulia kicks the spacebar, and chnges windows like she meant to13:34
*** lblanchard has joined #tripleo13:36
gfidenteEmilienM, I started with puppet-openstack because that is what astapor uses but I'm fine switching, let me update the submissions13:37
EmilienMgfidente: I should have told you, apologize13:38
gfidenteEmilienM, I will check with jayg if there are downsides, he might know better than me13:38
gfidenteEmilienM, no worries the changes for now are just meant to put in place bare minimum so it's big deal13:38
gfidente*it's not a big deal13:39
EmilienMgfidente: check with jpena also, he's very aware about that13:39
*** panda|afk is now known as panda13:46
*** panda is now known as Guest4497913:46
*** Guest44979 is now known as panda13:47
jistrgfidente, spredzy: i loopback-mounted the image to verify that there's no *-target.wants entry mentioning rabbitmq, and indeed there is not, so rabbitmq is probably not running before it's configured, it gets started by puppet13:49
gfidentejistr, check we might want to switch to puppet-corosync though13:50
gfidentesorry, I meant check
spredzygfidente, I checked that + a rpm -q --changelog on the node13:51
jistrgfidente, spredzy: then i forced the rabbitmq erlang cookie creation to happen before the service is started, which it did, but it still didn't fix the issue13:51
* jistr looks at the link13:51
spredzyit seems that all the fixed has been put int13:51
*** pradeep has quit IRC13:51
spredzythey do use Type=notify in the latest version of the package13:51
spredzy(ie. what apevec says later on the issue | cf. bz linked by jistr)13:52
openstackgerritMerged openstack/tuskar-ui: Move performance box code into a js file.
jistrgfidente: yeah i've seen that code. I think they put that in because they want the cookie to exist before the service is started. We create the cookie manually, but the effect should be the same -- the cookie is present at the time RabbitMQ starts for the first time.13:53
gfidentethe module deletes it and then starts; right ;(13:53
gfidentewell recreates it as well13:53
jistryeah but that should be ok, as at that point RabbitMQ is not running yet13:54
jistrat least when i added this for testing: Rabbitmq_erlang_cookie<||> -> Service['rabbitmq-server']13:54
jistrbut it still doesn't fix it13:54
openstackgerritMerged openstack/tuskar-ui: Handle traceback when Node is locked
jistri'm thinking we hit a different issue than the one eck hit back then. In his logs there's no mention of tables_not_present, even though the symptoms are very similar.13:55
jistr(so the BZ probably won't be a clue for us, we already do what it was supposed to fix)13:56
*** dtantsur|brb is now known as dtantsur13:56
*** sdake has joined #tripleo13:59
*** sdake_ has quit IRC14:02
*** pradeep has joined #tripleo14:02
*** stendulker has joined #tripleo14:04
*** rhallisey has quit IRC14:04
*** rhallisey has joined #tripleo14:06
gfidentejistr, do you know if corosync in staypuft uses multicast or broadcast?14:08
jistrgfidente: i don't, but this might be helpful maybe?
*** jerryz has joined #tripleo14:10
gfidentetks :)14:12
*** derekh changes topic to "CDeploying OpenStack Using OpenStack | Summit proposed topics: |"14:14
*** derekh changes topic to "Deploying OpenStack Using OpenStack | Summit proposed topics: |"14:14
derekhI see the puppet job is failing today? IS anybody looking into it ?14:16
*** pdb has quit IRC14:18
*** sdake_ has joined #tripleo14:18
*** pdb has joined #tripleo14:18
gfidentejistr, looks like known issue?
gfidentejistr, sorry,
*** BadCub_Hiding is now known as BadCub14:20
gfidentederekh, are we pinned to some t-h-t older than ?14:21
spredzygfidente, derekh this doesn't seem to be the issue as MariaDB now starts well14:21
*** sdake has quit IRC14:21
spredzyif you look a console.html the process goes all the way to the end, heat resource-list show everything as CREATE_COMPLETE14:23
gfidentespredzy, which job is this? not the one I posted the logs I think14:23
*** regebro has joined #tripleo14:23
spredzyduring destroy_vms it seems to fail. During pip-freeze it crashes with ImportError: cannot import name UnsupportedWheel. Not sure if that might be the issue14:24
spredzygfidente, not specific jobs, just all the recheck from 2/3 hours ago fails w/ no apparent reason (not apparent to me)14:24
derekhgfidente: nope we have no pins, we have a cherrypick but it should be doing nothing, I'll remove it now14:26
jistr2015-04-13 14:13:09.756 | OUTPUT=PING ( 56(84) bytes of data.14:27
jistr2015-04-13 14:13:09.756 | From icmp_seq=1 Destination Host Unreachable14:27
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: Remove heat-templates cherrypick
jistrsorry, this too:14:28
jistr2015-04-13 14:13:09.756 | Timing out after 300 seconds:14:28
jistr2015-04-13 14:13:09.756 | COMMAND=ping -c 1
gfidentespredzy, yeah the log I posted is older14:28
jistrlooks like it cannot ping the instance spawned in overcloud14:28
*** jistr is now known as jistr|mtg14:29
mordredgreghaynes: ping when you're awake14:30
*** dsneddon has joined #tripleo14:31
gfidentederekh, I don't see any compute log file in the compute nodes?14:35
derekhgfidente: which job are you looking at ?14:35
gfidentethe two nodes from same job14:36
derekhgfidente: it mustn't have gotten stated, will take a look in about 5 minutes14:37
*** jprovazn has quit IRC14:37
openstackgerritSteven Dake proposed stackforge/kolla: Use cento base image for data volumes
mordredgreghaynes: your latest changse to the simple-init patch do not seem to  have resulted in glean actually being installed in the image14:42
*** sdake has joined #tripleo14:47
*** rbrady has quit IRC14:49
*** pdb has quit IRC14:50
*** sdake_ has quit IRC14:50
*** rwsu has quit IRC14:54
*** rwsu has joined #tripleo14:54
*** jistr|mtg is now known as jistr15:07
*** jerryz has quit IRC15:07
*** sdake_ has joined #tripleo15:07
*** stendulker has quit IRC15:12
*** sdake has quit IRC15:13
bnemecgfidente: derekh: Puppet claims it was started, although I have recently discovered that doesn't mean much.15:14
bnemecAt least on the particular log I was looking at.15:14
bnemecNotice: /Stage[main]/Nova::Compute/Nova::Generic_service[compute]/Service[nova-compute]/ensure: ensure changed 'stopped' to 'running'15:17
gfidentebnemec, hostinfo log file says it was running indeed15:17
*** rbrady has joined #tripleo15:18
*** jprovazn has joined #tripleo15:19
bnemecLies! :-D15:20
derekhpr 13 14:58:43 ov-ivrwbayvn7-0-bqk7hj4ytdro-novacompute-blprjeps6qmc systemd[1]: Starting OpenStack Nova Compute Server...15:20
derekhApr 13 14:58:44 ov-ivrwbayvn7-0-bqk7hj4ytdro-novacompute-blprjeps6qmc systemd[1]: Started OpenStack Nova Compute Server.15:20
bnemecOkay, so maybe that's a red herring.15:20
*** pdb has joined #tripleo15:20
derekhHas anybody tried this locally ? going to kick it off now15:22
bnemecI haven't yet, but I was thinking along the same lines.15:22
bnemecOh look, a nova-compute log: InvalidVolume: Invalid volume: volume '025161e5-3939-408f-a3ae-7d336451a89b' status must be 'available'. Currently in 'error'15:23
bnemecSo it's a cinder problem15:23
bnemecFailed to copy image to volume: qemu-img: error writing zeroes at sector 0: Invalid argument15:24
*** yamahata has joined #tripleo15:30
gfidentederekh, bnemec I don't have it locally though with updates repos and images15:34
bnemecgfidente: derekh: Bleh: bug 1194743 in openstack-cinder "cinder: cannot create a volume from image" [Urgent,New] - Assigned to eharney15:34
bnemecI also see dprince commented on bug 1196688 in qemu-kvm-rhev "bdrv_make_zero() passes a too large nb_sectors value to bdrv_write_zeroes()" [Unspecified,Modified] - Assigned to stefanha15:34
dprincebnemec: a new version of qemu-img was released for F21 which should resolve that issue15:36
gfidentedprince, that might explain why I don't have it, do you know which version fixes it?15:37
bnemecLooks like we have qemu-img-2.1.3-5.fc21.x86_64 on the failing CI instance.15:37
gfidentedprince, I'm on f20, qemu-img-1.6.2-13.fc20.x86_6415:38
dprinceqemu-2.1.3-3.fc21 had it working15:39
bnemecOoh, we may have a regression.  The last puppet job passed had qemu-img-2.1.3-3.fc21.x86_6415:39
dprinceThis is the fedora ticket: bug 1200043 in qemu "bdrv_make_zero() passes a too large nb_sectors value to bdrv_write_zeroes()" [Unspecified,Closed: errata] - Assigned to virt-maint15:39
dprincebnemec: perhaps someone rebased... and broke it again15:39
*** mmagr is now known as mmagr|afk15:39
dprincebnemec: we should ping hguemar15:40
derekhbodhi - 2015-04-13 07:05:2115:41
derekhThis update has been pushed to stable15:41
*** spredzy is now known as spredzy|afk15:42
derekhbnemec: dprince gfidente did that cause it ^^15:42
bnemecderekh: I think that's likely.  The puppet jobs didn't start failing until today.15:42
bnemecdprince: You've got the pinging covered?15:43
dprincebnemec: um, sure. I guess I can take care of this again15:43
bnemecdprince: I can do it too, I just don't know where hguemar hangs out. :-)15:44
dprincebnemec: either way. The "Steps to reproduce" in the original ticket will tell you for sure if it is broken15:44
bnemecif /me were better at IRC he could probably figure that out.15:44
bnemecdprince: Okay, will try that first then.15:45
*** jistr has quit IRC15:45
bnemecHrm, my f21 vm doesn't even see -515:46
bnemecqemu-img                      x86_64     2:2.1.3-3.fc21      updates     585 k15:47
dprincebnemec: probably hasn't propigated to all the mirrors yet15:47
bnemecInstall from bodhi it is, then15:48
*** yamahata has quit IRC15:49
bnemecWe have a winner:15:50
bnemec[root@openstack tmp]# qemu-img convert -t none -O raw input.img /dev/loop015:50
bnemecqemu-img: error writing zeroes at sector 0: Invalid argument15:50
dprincebnemec: yeah, that is busted then15:51
dprincebnemec, derekh: in the meantime we could push an element fix to pin to the -3 version (which worked fine)15:52
* dprince doesn't mind if our CI jobs don't have the security patch for a day or so15:52
openstackgerritVictor Lowther proposed openstack/os-cloud-config: Add pxe_drac driver support to os-cloud-config
derekhdprince: yup agreed, a fix will take a few days to get out to fedora repositories15:53
*** alop has joined #tripleo15:57
openstackgerritBrad P. Crochet proposed openstack/tripleo-puppet-elements: Add newline to end of source-repository-puppet-modules
*** cwolferh has joined #tripleo16:02
*** mmagr|afk is now known as mmagr16:02
*** ifarkas has quit IRC16:07
dprincederekh: I don't think we keep the old one around in the repo though16:08
dprincederekh: I'm unable to pin to it :(16:08
dprincederekh: so short of stashing our own copy somewhere it is just a bust16:09
derekhdprince: would this work ? yum install -y --nogpg
dprincederekh: probably, would need to do it in an element though16:14
dprincederekh: -p for DIB wouldn't get you that16:14
derekhdprince: yup, I thought that was what you were trying todo16:14
dprincederekh: we can use an element, sure16:15
dprincederekh: I was trying to use -p initially to keep this lighter16:15
dprincederekh: as in just a patch to tripleo-ci16:15
derekhdprince: ack16:16
*** eghobo has joined #tripleo16:22
*** Goneri has quit IRC16:22
*** eghobo has quit IRC16:23
openstackgerritDan Prince proposed openstack/tripleo-image-elements: Add an element to pin to Qemu 2.1.3-3 on Fedora 21
*** lsmola_ has quit IRC16:25
openstackgerritDan Prince proposed openstack/tripleo-image-elements: Add an element to pin to Qemu 2.1.3-3 on Fedora 21
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: Pin qemu-img on Fedora 21 jobs.
dprincederekh: ^^16:28
derekhdprince: looking16:28
dprincederekh: no need to land the element itself. I figured we could just cherry pick it16:28
derekhdprince: +216:30
* derekh isn't sure if you'#ll need the --nogpg or not16:30
*** pblaho has quit IRC16:33
*** ukalifon1 has quit IRC16:34
*** panda is now known as panda|afk16:36
dprincederekh: those packages should be signed16:36
derekhdprince: k]16:37
derekhdprince: k16:37
*** subscope_ has joined #tripleo16:41
*** mmagr has quit IRC16:42
*** mmagr has joined #tripleo16:43
*** eghobo has joined #tripleo16:50
openstackgerritMerged stackforge/kolla: Adds neutron-server to neutron container-set
openstackgerritMerged stackforge/kolla: Adds Neutron Agents Support
*** stendulker has joined #tripleo16:55
*** yamahata has joined #tripleo16:57
*** derekh has quit IRC16:57
*** dtantsur is now known as dtantsur|afk16:57
*** stendulker has quit IRC16:59
*** wfoster has quit IRC17:02
*** Marga_ has quit IRC17:04
*** bvandenh has quit IRC17:04
*** Marga_ has joined #tripleo17:05
*** Marga_ has quit IRC17:05
*** Marga_ has joined #tripleo17:06
greghaynesmordred: ohai17:07
greghaynesmordred: hrm17:07
*** subscope_ has quit IRC17:08
*** wfoster has joined #tripleo17:09
*** eghobo_ has joined #tripleo17:11
*** eghobo has quit IRC17:14
*** eghobo has joined #tripleo17:15
mordredgreghaynes: if I hack in a pip install glean in an install script it works as expected17:15
mordredbut if I don't - it doesn't17:15
mordredgreghaynes: btw - I just got accoutns on two new clouds17:15
mordredgreghaynes: and in a fit of insanity - so far everything we've done just works there17:15
greghayneshaha, awesome17:16
greghaynesim messing with the glean install now17:16
mordredI think that's the last remaining piece - the other stuff is looking really solid17:16
mordredgreghaynes: also - I pushed up SEVERAL patches to glean from things I found live on the new cloud I poked at17:16
mordredgreghaynes: mostly related to path roots and stuff17:17
greghayneswhat clouds?17:17
greghaynesalso, sample data?17:17
*** eghobo_ has quit IRC17:17
openstackgerritDaneyon Hansen proposed stackforge/kolla: Removes check_required_vars for Log Files
bnemecdprince: Looks like the pin patch failed: :-(17:19
dprincebnemec: yeah, I'm working on a modification.17:19
dprincebnemec: it installed the right packages and then our 'updates' script updated it17:20
dprincebnemec: I'm trying an 'exclude=' to the yum file now17:20
bnemecdprince: Cool, thanks17:20
bnemecI still think we should move the update step to be first instead of last.17:21
bnemecWe might as well install current versions of all the things in the first place.17:21
dprincebnemec: that would have made this a bit more intuitive for sure17:21
bnemecI might just propose it.  We've been bitten by this sort of thing a few other times too.17:23
openstackgerritSteven Dake proposed stackforge/kolla: Update developer documentation
*** jcoufal has quit IRC17:26
*** jsidhu has joined #tripleo17:33
*** Marga_ has quit IRC17:33
*** tzumainn has quit IRC17:37
*** Marga_ has joined #tripleo17:40
*** trown is now known as trown|lunch17:41
*** mmagr has quit IRC17:41
*** Marga_ has quit IRC17:44
openstackgerritSteven Dake proposed stackforge/kolla: Update developer documentation
openstackgerritDan Prince proposed openstack/diskimage-builder: Update packages earlier
*** jsidhu has quit IRC17:53
*** athomas has quit IRC17:55
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: Pin qemu-img on Fedora 21 jobs.
openstackgerritBen Nemec proposed openstack/diskimage-builder: Update installed packages first instead of last
dprincebnemec: I stole your idea :)17:55
dprincebnemec: Sorry. I actually couldn't seem to get it working without it17:56
dprincebnemec: rather than fight with it further I just updated the patch to cherry pick it17:56
bnemecdprince: No problem, as long as it happens somehow I'm happy. :-)17:56
dprincebnemec: with both of those I think we should see it passing...17:56
dprincebnemec: yeah, sorry. It is like 2 hours later and I'm just anxious to get back to working on what I was doing to begin with17:57
*** mgagne_ is now known as mgagne17:58
bnemecdprince: I was actually going to suggest that you might want to just rebase on that patch anyway, so it makes perfect sense to me.17:58
dprincebnemec: you used 00 and I used 0117:58
dprincebnemec: is one better?17:59
dprincebnemec: pretty much the same effect I think17:59
bnemecdprince: I dunno, I was just trying to get it done as early as possible, but it probably doesn't hurt to leave some way to run a hook before the update happens either.17:59
*** pelix has quit IRC18:00
*** eghobo has quit IRC18:03
*** eghobo has joined #tripleo18:04
mordredgreghaynes: the fixtures - they were in the fixtures dir at a mix of rooted on / and rooted on /mnt/config18:08
mordredgreghaynes: I moved them all to be rooted on / so that the things supposed to be in /mnt/config are in fixtures/mnt/config18:08
mordredgreghaynes: it was getting us into a place where we were happily passing tests but failing ssh key injections on a real box because it was looking for /mnt/config/root/.ssh/authorized_keys18:09
*** Marga_ has joined #tripleo18:09
mordredgreghaynes: all works now - verified on new clouds18:09
greghaynesah, yea, I remember doing some overly zealous monkeypatching for that18:09
mordredgreghaynes: also - the new clouds are vexxhost and runabove18:09
mordredgreghaynes: I have mildly twiddled your monkeypatching. in general it's all good18:09
mordredgreghaynes: I think the errors may have come in the move into the new repo and corresponding path change18:10
openstackgerritBrad P. Crochet proposed openstack/python-tuskarclient: Correct plan-remove-role to handle return of delete correctly
*** Marga_ has quit IRC18:12
*** Marga_ has joined #tripleo18:13
*** Marga_ has quit IRC18:13
*** Marga_ has joined #tripleo18:13
*** eghobo has quit IRC18:16
*** jsidhu has joined #tripleo18:18
*** rpodolyaka has quit IRC18:22
*** trown|lunch is now known as trown18:26
*** lucasagomes is now known as lucas-dinner18:30
*** SlickNik has quit IRC18:31
openstackgerritgreghaynes proposed openstack/diskimage-builder: Add element to process config-drive network info
*** SlickNik has joined #tripleo18:32
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: Pin qemu-img on Fedora 21 jobs.
*** Marga_ has quit IRC18:38
greghaynes^ wow18:38
*** Marga_ has joined #tripleo18:38
*** dprince has quit IRC18:43
*** Goneri has joined #tripleo18:44
openstackgerritMerged stackforge/kolla: Removes check_required_vars for Log Files
openstackgerritSteven Dake proposed stackforge/kolla: neutron-agents has wrong yml syntax
mordredgreghaynes: didyou find it?18:53
*** regebro has quit IRC18:54
mordredgreghaynes: HAHAHAHAHAHAHAHA18:54
mordredgreghaynes: NICE18:54
greghaynesmordred: AFAICT by default it did install, the repo installtype was broken thouhg18:54
greghayneswere you seeing boot failures?18:54
openstackgerritDaneyon Hansen proposed stackforge/kolla: Fixes Spacing in Neutron Agents Compose File
mordredgreghaynes: I was seeing glean not in /usr/local/bin18:54
mordredgreghaynes: which I verified by logging in to the host and seeing that it was not there18:55
greghaynesim booting up an image now so ill see18:55
mordredgreghaynes: I did not set anyoptions in terms of source vs. package18:55
greghaynesits a dibism - if you export DIB_INSTALLTYPE_simple_init=repo itll install from git repo rather than pip now18:55
mordredbut if I do nothing it should install from pip, yeah?18:56
mordredbtw - can we make an element that installs pip not from python-pip ?18:56
mordredbut does get-pip instead?18:56
mordredor does that already exist?18:56
greghaynesyes - we have that in tripleo, it should get moved into dib18:56
mordredit's important - python-pip is the devil in my opinion18:56
mordredit provides no value except for breaking things18:57
greghaynesthatll be an interesting dependency problem18:57
greghaynestheres not a good way for a user to opt-in to the get-pip element18:58
*** dprince has joined #tripleo18:59
openstackgerritSteven Dake proposed stackforge/kolla: Prepare compose for custom namespace
greghaynesI see whats going ong19:07
greghaynesenable install types is in the base element19:07
greghaynesand on my testing box I wasnt using -n19:10
openstackgerritMerged stackforge/kolla: Fixes Spacing in Neutron Agents Compose File
openstackgerritgreghaynes proposed openstack/diskimage-builder: Break install-types out of base
SpamapSgreghaynes: yeah I tried once to do that and got told no. But maybe now we have two reasons. :)19:38
greghayneswhy no19:39
openstackgerritDaneyon Hansen proposed stackforge/kolla: Changes Default Image Prefix
openstackgerritgreghaynes proposed openstack/diskimage-builder: Add element to process config-drive network info
greghaynesmordred: ok, ^ should *actually* fix it19:41
*** regebro has joined #tripleo19:41
openstackgerritgreghaynes proposed openstack/diskimage-builder: Add a yum-minimal element that just uses yum
openstackgerritgreghaynes proposed openstack/diskimage-builder: Port centos-minimal to yum-minimal
mordredgreghaynes: is the fix rebasing on top of install-types?19:45
greghaynesand adding that as a dep19:46
mordredgreghaynes: ossum!19:46
mordredSpamapS: what did you get told no to? breaking install-types out? or get-pip?19:46
greghaynesbreaking install-types out I bet19:46
SpamapSmordred: breaking install-types out.19:47
mordredoh - well - I mean, I could understand not if base was actually base19:47
SpamapSit had only just been put in base, so I think we were in a different place.19:47
* mordred keeps meaning to take a hatchet to base more19:47
*** akrivoka has quit IRC19:49
openstackgerritRyan Brady proposed openstack/tuskar-ui: Fix node list
*** sdake has joined #tripleo19:53
*** Marga_ has quit IRC19:56
openstackgerritBen Nemec proposed openstack/tripleo-image-elements: Add an element to pin to Qemu 2.1.3-3 on Fedora 21
*** sdake_ has quit IRC19:56
openstackgerritBen Nemec proposed openstack-infra/tripleo-ci: Pin qemu-img on Fedora 21 jobs.
dprincebnemec: sigh, one of those days20:04
dprincebnemec: thanks for bumping the refspec...20:05
greghaynesdprince: Hey, replied on
bnemecdprince: This is pretty much BAU for me lately. :-)20:06
dprincegreghaynes: having to pin packages is totally not intuitive20:06
bnemecAlso, there are other problems we've had to work around in the past that would have been fixed by updating first.20:07
greghaynesyes, I just am afraid its not sustainable to only update at the right times - we want to do updates at the end up pre-install so we can take in things like added repos20:07
bnemecAlso also, I don't know that there is a single installer out there that tells you to start with an outdated system.20:07
dprincegreghaynes: this is an install.d taks though20:07
dprincegreghaynes: repo files should generally be done in pre-install.d, and we update via install.d20:08
dprincegreghaynes: that is what happens now anyways20:08
greghayneshuh, ubuntu does pre-install/99- for that but close enough20:08
greghaynesthe issue is we do install some things in pre-install20:08
bnemecI'm not too concerned about added repos anyway.  They'll be handled the first time something gets installed from them.20:09
greghaynesso if we want it to work, I think itll end up having to go at 00- in pre-install, and then you have a land mine of hoping nowhere else does an update (like the update that will be done to add repos)20:10
*** lucas-dinner has quit IRC20:10
dprincegreghaynes: still not sure I follow your reasoning20:10
dprincegreghaynes: it does appear to work, CI passes20:11
* greghaynes digs for where we install qemu-img20:12
dprincegreghaynes: Nova installs it20:12
bnemecAdding a repo isn't going to call yum update or apt-get upgrade though.  They'll pull down new package lists, but unless you tell them to update the installed packages they shouldn't.20:12
bnemecIt looks like it's coming from a package-installs file.20:13
*** rpodolyaka has joined #tripleo20:13
greghaynesah, no, its part of an install.d script20:14
greghaynesok - my issue is this only works for a small subset of packages right now - itll only work for things that are past install.d/01- (which package-installs is not)20:14
greghaynesso it doesnt actually make pinning 'work'20:15
greghaynesI dont believe there is a sane fix for that though, we can try bumping it forward or moving package-installs back but itll still have a big caveat20:15
bnemecHonestly I'm not too concerned about handling all of the edge cases.  My main thing is that it makes more sense to start out with an updated environment and then do our installs.20:17
bnemecThis pinning thing was kind of a happy side-effect of that.20:18
greghaynesYep, thats fair - so for that id be fine with a 0something in pre-install.d20:18
greghaynesWhen you say pinning is very non-intuitive, are you just concerned that its a pain?20:18
greghaynesor is there some techinical issue with it20:19
bnemecOh ffs, now it blew up on downloading the image.20:20
* bnemec didn't want to have working CI again anyway.20:20
bnemecTBH, I have no idea how pinning works in anything but Gentoo.  I believe Dan started looking into it for this and decided it wasn't worth the hassle, but I don't have more detail than that.20:22
* dprince will be back in a bit20:22
*** Marga_ has joined #tripleo20:24
*** lblanchard has quit IRC20:24
dprincegreghaynes: there are many ways to pin packages20:24
dprincegreghaynes: all of them should work so long as they come after 0120:24
dprincegreghaynes: I sort of feel like putting it into pre-install is a violation of what "pre-install" is supposed to be20:25
dprincegreghaynes: so how about 00 instead? which is what bnemec did initially anyways...20:25
bnemecdprince: If we do that, then we need to move the package-installs script from 00 to 01 or we have a race.20:26
bnemecWe could just make it come first alphabetically, but I think we officially don't support ordering within numerical steps like that.20:26
dprincebnemec: fine w/ me20:26
greghayneswhy not for pinning just install the specific version you want early, add it to yum.conf excludes, then it doesnt matter if anything runs an update later on?20:27
bnemecMe too, just making a note.20:27
*** eghobo has joined #tripleo20:27
dprincegreghaynes: because it that isn't always the way we get packages. SEe:
dprincegreghaynes: there are many ways to get an RPM, and a Yum update at a later time could blow them away20:28
dprincegreghaynes: which is why updating earlier makes this more intuitive20:28
greghaynesoh, so an exclude on qemu-img doesnt work for an rpm installed that wa?20:28
*** eghobo_ has joined #tripleo20:29
dprincegreghaynes: I'd have to pin the fedora-updates.repo file20:29
dprincegreghaynes: which honestly I was messing with (via exclude=qemu-img, etc) but it wasn't working20:29
*** Goneri has quit IRC20:29
dprincegreghaynes: and wasn't intuitive either20:29
dprincegreghaynes: ad-hoc elements should be easy20:29
greghaynesyea, ok. So then another suggestion if were going this route - the dpkg element currently does an apt-get update in pre-install.d/99-, we should use that20:30
*** bnemec changes topic to "CI failing due to regression of | Deploying OpenStack Using OpenStack | Summit proposed topics: |"20:31
greghaynesotherwise ubuntu is going to do an update in pre-install/99 then again in install/00-20:31
* greghaynes removes -220:31
openstackgerritDaneyon Hansen proposed stackforge/kolla: Sets Neutron log_dir=/var/log/neutron as default
*** eghobo has quit IRC20:32
greghaynesor maybe just remove the dpkg one along with this20:33
*** HenryG has quit IRC20:33
bnemecgreghaynes: So basically ubuntu is already doing this?  Except on the other side of the pre/install.d split?20:33
greghaynesit does it because elements need to be able to add repos in pre-install.d20:34
greghaynesand then we want to auto update the package index before install.d20:34
openstackgerritDaneyon Hansen proposed stackforge/kolla: Sets debug logging to false
dprincegreghaynes: that is a different behavior, apt-get update isn't the same as yum update20:35
bnemecYeah, but I think running the up-to-date script would accomplish the same thing.20:36
* bnemec hasn't actually looked at the up-to-date script though.20:36
bnemecDetails. :-)20:36
dprincegreghaynes: I still have a slight preference for this going into install.d, but if we must converge the two a pre-install.d/99 may be fine too (or close enough)20:36
*** HenryG has joined #tripleo20:37
dprincebnemec: regardless of what we do with this DIB branch lets go ahead and land the tripleo-ci cherrypicks as soon as we see the Fedora puppet job passing20:38
openstackgerritMerged stackforge/kolla: Sets Neutron log_dir=/var/log/neutron as default
greghaynesdprince: thats fine with me - we should guard apt distros out of it then or something along those lines20:39
bnemecdprince: Yeah, I was planning to.  The stupid puppet job failed downloading the base image for some reason though. :-(20:39
greghaynesalso, right20:39
* greghaynes was conflating update and upgrade20:39
bnemecWell, they mean different things in apt and yum.  Because consistency is for other people. :-)20:39
dprincebnemec: right, but it looks like a new one is running now right?20:41
greghaynesso, now that im not confusing the two (I think) im actually wrong about needing to guard ubuntu out of the change youre making20:41
*** radez is now known as radez_g0n320:41
bnemecdprince: It's the new one that failed:
dprincebnemec: oh, I was looking at the wrong patch20:43
bnemecI need to duck out for a while to drop my truck off for some work.20:43
dprincebnemec: cool, thanks for pushing this along. ttl20:44
greghaynesim now wondering why we even need the 99-update at all20:44
bnemecHopefully I'll be back soon, but history suggests it's kind of a crapshoot. :-)20:44
openstackgerritMerged stackforge/kolla: Changes Default Image Prefix
greghaynesSomehow doing a dist-upgrade prevents us from needing to install linux-image-generic?20:46
greghayneslifeless: ^ do you remember how/why?20:48
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: Pin qemu-img on Fedora 21 jobs.
*** sdake_ has joined #tripleo20:53
*** sdake has quit IRC20:57
*** morazi has quit IRC21:03
*** trown is now known as trown|outttypeww21:04
*** Marga_ has quit IRC21:09
*** Marga_ has joined #tripleo21:10
*** jprovazn has quit IRC21:11
lifelessgreghaynes: context?21:13
greghaynesTrying to figure out why we need a 99-up-to-date script, seems its to get the linux-image-generic package?21:14
lifelessgreghaynes: IIRC - we were installing linux-image-generic by hand not to install it, but to ensure we had the latest version21:15
greghaynesah, so if we got a cloudimg with an old kernel21:15
*** Goneri has joined #tripleo21:16
lifelessremembering its a metapackage21:17
lifelessso it depends on current-kernel-versioned-package,in ubuntu21:17
lifelessthere isn't that metapackage in ARM-hf21:17
lifelessbecause older iterations of engineers hated their future selves21:17
lifelessgreghaynes: so, apt-get dist-upgrade {the dist- is important} gets the new metapackage and the new kernel on x86 and amd64, and the new whatever metapackage the arm image in use has, and *its* newer kernel21:23
* bnemec is back, at least for another 30-45 minutes21:29
greghaynesbnemec: so, im fine with just pushing package-installs back to 01-21:30
greghaynesif thats the path you want to go21:30
bnemecgreghaynes: Cool, I think that works for me.  It doesn't necessarily solve all of the pinning cases, but it's enough to make relatively simple ones like this work.21:37
bnemecAnd like I said, that wasn't my primary motivation for suggesting this in the first place.21:37
*** regebro has quit IRC21:37
*** regebro has joined #tripleo21:38
greghaynesbnemec: yep, and actually I think itll have the same effects - things we install ontop of the base image shouldnt be upgradeable21:39
greghaynessince only the base image contents can be out of date21:40
*** cody-somerville has joined #tripleo21:45
*** cody-somerville has joined #tripleo21:45
*** jtomasek has quit IRC21:46
openstackgerritBen Nemec proposed openstack/diskimage-builder: Update packages earlier
dprincebnemec: may as well add yourself as a co-author on this as it *was* your idea sir21:49
openstackgerritBen Nemec proposed openstack/diskimage-builder: Update packages earlier
bnemecdprince: Done.  Until I push another patch set and overwrite the inline Gerrit change. :-)21:50
*** Marga_ has quit IRC21:52
*** daneyon_ has joined #tripleo21:53
*** Marga_ has joined #tripleo21:56
*** daneyon has quit IRC21:56
*** Marga_ has quit IRC21:57
*** Goneri has quit IRC22:01
*** rhallisey has quit IRC22:13
SpamapSgreghaynes: Oh your statement about pushing package installs back to 01 reminded me of something22:37
SpamapSgreghaynes: I think we should seriously consider not _ever_ putting anything < 10 or > 9022:37
SpamapSgreghaynes: Strikes me that we're taking away users' ability to hackstend dib by putting things in those spaces.22:38
SpamapSIt's fine in collections of elements22:38
SpamapSbut not in dib itself22:38
greghaynesThey just need to have ;)22:39
SpamapSof course, this is where we go "damnit, why don't we have dependency expression?" and then we go "insserv!" and then we go "mmm beer" and forget.22:39
*** eghobo_ has quit IRC22:41
*** Marga_ has joined #tripleo22:49
*** daneyon has joined #tripleo22:56
*** dprince has quit IRC22:56
*** gfidente has quit IRC22:57
*** daneyon_ has quit IRC23:00
*** cody-somerville has quit IRC23:09
openstackgerritSteven Dake proposed stackforge/kolla: Make glance work reliably
*** chlong has joined #tripleo23:14
*** rlandy has quit IRC23:23
*** regebro has quit IRC23:27
*** cody-somerville has joined #tripleo23:27
*** jsidhu has quit IRC23:33
*** daneyon has quit IRC23:38
*** yuanying has joined #tripleo23:39
*** weshay has quit IRC23:52
*** tzumainn has joined #tripleo23:52

Generated by 2.14.0 by Marius Gedminas - find it at!