Friday, 2014-10-17

*** yuanying_ has joined #tripleo00:01
*** ramishra has quit IRC00:02
lifelesstchaypo: win!00:02
*** yuanyin__ has joined #tripleo00:04
*** yuanying has quit IRC00:04
*** yuanying_ has quit IRC00:06
tchaypoand so finally i can ask register-nodes to update my list of nodes00:06
tchaypoand I get00:06
tchaypoBadRequest: Invalid input for field/attribute value. Value: '49152'. Wrong type. Expected '<type 'unicode'>', got '<type 'int'>' (HTTP 400)00:06
lifelesstchaypo: yay, you can give StevenK a bug report :)00:11
*** rainya has quit IRC00:11
lifelessSpamapS: whats this for?00:11
SpamapSlifeless: ansible00:12
tchaypolifeless: nope, his docs make it clear that the memory should be a string00:12
lifelesstchaypo: right, but was it a string before?00:12
tchaypookay that's a lie, his docs are silent. His *sample* uses a string00:12
lifelesstchaypo: and was it working before ?00:12
lifelesstchaypo: if you have a working input file for the old register-nodes, that makes the new one incompatible00:13
SpamapSlifeless: what we have right now is basically a monolithic workflow+tiny orchestration bits. What we want is the orchestration bits hidden safely behind Ansible's built in abstraction which has a terrible name, "roles"00:15
tchaypoMy file has changed since I last ran the old register-nodes00:15
tchaypo is the docs in -incubator for the old register-nodes and it says that the memory is in KB00:15
tchaypo(no mention of string vs int)00:15
tchaypo(see step 6)00:16
tchaypo is the new register-nodes; it just has "2048" with no mention of what the unit is.00:16
tchaypoI'm fairly certain StevenK isn't suggesting deploying nodes with 2048k of ram though.00:16
tchaypoit looks like the old register-nodes had:00:18
tchaypoironic node-update $NODE_ID add $(jq -r '"properties/cpus=" + (.cpu|tostring) + " properties/memory_mb=" + (.memory|tostring) + " properties/local_gb=" + (.disk|tostring) + " properties/cpu_arch=" + .arch' <<< $NODE_JSON)00:18
tchayposo I'm guessing it would take either a string or an int..00:18
tchaypoand testing confirms it.00:23
*** otoolee has quit IRC00:29
uvirtbotLaunchpad bug 1382275 in os-cloud-config "new register-nodes does not accept ints for numeric input" [Critical,Triaged]00:30
*** noslzzp has joined #tripleo00:33
openstackgerritA change was merged to stackforge/kolla: configure nova database connection
openstackgerritA change was merged to stackforge/kolla: Run nova-api as privileged so it can muck with iptables
*** ChuckC has quit IRC00:48
tchaypo... and now I discover that register_nodes will abort if it hits one node that's locked by nova, instead of skipping it and moving on00:49
tchaypoand the error message just tells me there was a "Conflict"00:49
tchaypo(more info is available in debug, but I shouldn't need to debug)00:49
lifelesstchaypo: bug #2 ;)00:50
*** marun has quit IRC00:52
*** ChuckC has joined #tripleo00:54
*** noslzzp has quit IRC00:56
cineramabug party!00:56
*** ramishra has joined #tripleo00:58
*** jerryz has joined #tripleo00:59
*** yuanyin__ has quit IRC01:00
tchaypono, I think that's #2 *and* #301:00
tchaypo#2 is the fact that the default output doesn't tell me what's happening01:00
tchaypobug #3 is the fact that the error message doesn't tell me what the error was01:00
uvirtbotLaunchpad bug 3 in launchpad "Custom information for each translation team" [Low,Fix released]
SpamapShm.. setting up a new seed-only.. seed just sits at "booting from hard disk" and then something about text being deprecated01:00
tchaypoand I guess bug #4 is the fact that it shouldn't abort when it hits a single locked node01:01
uvirtbotLaunchpad bug 4 in launchpad "Importing finished po doesn't change progressbar" [Medium,Fix released]
tchaypothanks uvirtbot01:01
*** RioS2 has joined #tripleo01:01
*** ramishra has quit IRC01:02
*** yuanying has joined #tripleo01:03
jerryzSpamapS: if i want to make it HA mode, what should i do? just assign 3 controllers?01:03
SpamapSjerryz: CONTROLSCALE=301:04
SpamapSjerryz: sorry, OVERCLOUD_CONTROLSCALE=301:04
*** rainya has joined #tripleo01:04
cineramaman, bashate really doesn't like "## " does it01:04
jerryzSpamapS: endpoints url will be haproxy IP automatically?01:05
*** yuanying has quit IRC01:05
*** yuanying_ has joined #tripleo01:05
SpamapSjerryz: correct01:08
SpamapSjerryz: a VIP will be allocated from the range you allocate to the overcloud, and that will be your API endpoint for everything.01:10
*** otoolee has joined #tripleo01:11
jerryzSpamapS: that VIP need to be in the same range as all API servers?01:12
jerryzSpamapS: can i use public IP for front ends and an internal for backends?01:14
SpamapSjerryz: that is theoretically possible, but I don't think devtest supports that.01:14
SpamapSactually hmm01:15
SpamapSit might01:15
SpamapSjerryz: I think you can actually do that01:16
lifelesswe did the work for it for CI01:17
lifelessjust the patches haven't all landed01:17
lifelessits only been 4 months01:17
SpamapSThe templates allow setting a different one on the PublicVirtualIP.01:17
lifelessgnar, sorry, didn't mean to be bitchy01:17
SpamapSBut I know that just having the port doesn't mean we know how to use it :-P01:17
lifelesspublicvip is in the public network for the overcloud01:18
lifelessit has to be in the same public network01:18
lifelessthe backends don't live on the public network01:18
lifelessthey live in the controlplane network01:18
lifeless(and neither is the tenant network- it can be totally different)01:19
SpamapSlifeless: PublicVirtualNetwork defaults to ctlplane01:19
lifelessyes defaults01:19
lifelessthe vlan patches haven't all landed01:19
lifelessexamine the cd-cloud config for hp1 or hp201:19
lifelessand te tripleo-end-to-end config for their patchesets01:20
*** sdake_ has quit IRC01:20
SpamapSlifeless: ok, jerryz, btw, if you haven't figured it out, is our friend from Huawei :)01:20
lifelessSpamapS: I know :) - we've sat down at actual tables before01:20
SpamapSlifeless: and so is trying to stand up a TripleO cloud on 16 machines. :)01:20
SpamapSjust making sure01:20
SpamapSand with that I retire to family time for a bit01:23
* StevenK ponders a trip outside to reposition a chain saw01:23
cineramaStevenK: that sounds....ominous01:24
StevenKOh, now they've started a chipper01:24
StevenKWhich is louder, more annoying and harder to reposition01:24
jerryzSpamapS, lifeless: thank you. will figure it out.01:24
lifelessjerryz: you may want to talk to tchaypo to get a current tripleo-end-to-end yaml for this01:25
cineramais there a consensus opinion on one-liner bash if statements style-wise?01:25
jerryzlifeless: ok.01:25
StevenKPersonally, I like: if [ -n "$FOO" ]; then01:25
lifelesscinerama: such as?01:26
cineramalifeless: e.g. if [ $? != 0 ] ; then echo "Terminating..." >&2 ; exit 1 ; fi01:27
lifelesscinerama: I suspect enough folk will find it hard enough to read to want it multi-lined01:27
cineramabashate reports not-so-useful error messages in that case and in the case of01:27
cineramaif [ condition ]; then #nodocs01:28
openstackgerritRichard Su proposed a change to openstack/tripleo-image-elements: SELinux: Update keepalived custom policy
*** nosnos has joined #tripleo01:32
openstackgerritSteve Kowalik proposed a change to openstack/os-cloud-config: Deprecate PKI setup in init-keystone
openstackgerritSteve Kowalik proposed a change to openstack/os-cloud-config: user is no longer required for init-keystone
tchaypojerryz: actually my yaml is a bit out of date. I was just starting to look at it last night and I realised that the triple-incubator in particular requires a big rebase before I can use it.01:40
StevenKSpamapS: When you're back, perhaps you should resurrect the Kilo mid-cycle sprint e-mail so we can get some people interested in it01:41
cineramayay midcycle!01:43
jerryztchaypo: i can take apart, customize the steps, do it step by step if necessary01:44
StevenKOh good, they've driven off with the chipper01:46
StevenKNow I can stop contemplating murder01:46
cineramaI live near a sports bar with an outdoor area so I periodically hear cheering01:47
cineramai have no idea what they're watching, so I'm like "YAY SPORTS"01:47
StevenKgus: Every time your mail mentions "nsenter" I have to re-read it to work it out01:48
StevenK"N Senter? No, that doesn't work. Oh, maybe it's NS for namespace ..."01:49
gusya, namespaces.01:49
gusnobody talks about cgroups, which is the other half of "containers"01:50
greghaynesStevenK: Did you see your favorite of patches is back01:51
StevenKgreghaynes: ... no?01:52
StevenKLet me get the brain bleach ready so I can forget quickly01:53
StevenKgreghaynes: Is that a simple resurrection?01:53
greghaynesThe commit message has changed01:54
StevenKgreghaynes: You will answer for your sins (re: the t-i-e changes) in Paris01:55
greghaynesyea, its bad01:55
StevenKI really really hate mustache sometimes01:55
StevenKgreghaynes: However, I think your t-h-t changes need a rework01:56
StevenKThe t-i-e changes look okay, but I think I need to find caffeine or something stronger to parse the template.01:57
greghaynesah, thats fair01:57
StevenKgreghaynes: The undercloud is what kept going "BANG!" during my testing before the Revert of Doom01:57
*** rainya has quit IRC01:58
StevenKgreghaynes: Where's the toci patch up to?01:59
*** rainya has joined #tripleo01:59
*** ramishra has joined #tripleo01:59
greghayneswhich one?02:00
*** weshay has quit IRC02:00
greghaynes ?02:01
StevenKgreghaynes: Yup02:01
greghaynesI think it works02:01
StevenKgreghaynes: Yes, but you said the same thing about the ssl_pki branch :-P02:03
lifelessStevenK: revert of doom?02:03
greghaynesStevenK: haha, in all fairness the ssl pki patch landed while I was on PTO02:03
*** ramishra has quit IRC02:03
StevenKlifeless: ssl_pki was reverted out of os-cloud-config because it broke the world02:03
greghaynesalso I didnt realise that CI was a total lie for that repo :p02:03
greghaynesI wish there was some reasonably way to test that toci patch but pretty sure we just have to go for it02:04
StevenK% git show d1a22691865f999212097dff2d2ac49627479a1e | diffstat -s02:04
StevenK 8 files changed, 341 insertions(+), 495 deletions(-)02:04
StevenKTherefore, Revert of Doom02:04
openstackgerritgreghaynes proposed a change to openstack-infra/tripleo-ci: Build a local pip mirror of python packages being tested
StevenKgreghaynes: So you fixed the commit message to remove WIP: on the short line, but didn't remove the WIP bit at the end?02:06
openstackgerritgreghaynes proposed a change to openstack-infra/tripleo-ci: Build a local pip mirror of python packages being tested
greghaynesStevenK: huh? where?02:06
StevenKIn patch set 4, you cheater :-P02:06
StevenKgreghaynes: I think my only complaint about it is some of the comments need re-flowing02:08
greghaynesThe todos02:09
*** rainya has quit IRC02:09
StevenKgreghaynes: Here, have a proper review02:12
*** rainya has joined #tripleo02:20
*** rainya has quit IRC02:23
stevebakerSpamapS: ugh, having renamed the request collector to url, I now remember that heat has been released assuming it is called 'request'
*** rainya has joined #tripleo02:32
*** rhallisey has quit IRC02:34
*** pradeep has joined #tripleo02:35
*** pradeep1 has quit IRC02:35
jpeelersdake: ryan beat me. fyi, i usually am away from my computer 3 pm your time onwards (sometimes i return late like now).02:39
*** cwolferh_ has joined #tripleo02:41
*** spzala has quit IRC02:44
*** rainya has quit IRC02:46
jpeelerlarsks: i guess you missed this blog post -
larsksjpeeler: ??02:48
larsksOh, is it released?02:48
larsksAlso, --fixed-cidr is cool.02:48
jpeelerwhat does that do? (not a docker expert)02:49
*** rainya_ has joined #tripleo02:49
larsksSo, pre 1.3, docker would feel free to use the entire address space associated with the docker bridge -- so attaching containers directly to a local network was basically impossible, because docker would tromp on existing addresses.02:49
larsksWith 1.3, you can limit the address ranged used by docker with --fixed-cidr.02:49
jpeeleroh nifty02:50
*** rainya has joined #tripleo02:51
*** ramishra has joined #tripleo02:52
*** rainya_ has quit IRC02:52
*** ramishra_ has joined #tripleo02:58
*** untriaged-bot has joined #tripleo03:00
untriaged-botUntriaged bugs so far:03:00
*** untriaged-bot has quit IRC03:00
uvirtbotLaunchpad bug 1381408 in diskimage-builder ""visudo" always got error when building overcloud image" [Undecided,New]03:00
*** ramishra has quit IRC03:01
*** ramishra_ has quit IRC03:02
*** ramishra has joined #tripleo03:02
*** ramishra has quit IRC03:07
*** pradeep has quit IRC03:13
*** pradeep has joined #tripleo03:13
*** cwolferh_ has quit IRC03:15
*** yamahata has joined #tripleo03:18
*** ramishra has joined #tripleo03:22
*** julim has joined #tripleo03:24
openstackgerritJames Polley proposed a change to openstack/os-cloud-config: If one node is locked, keep trying the rest.
*** jcoufal has joined #tripleo03:27
*** pradeep1 has joined #tripleo03:30
*** pradeep has quit IRC03:32
*** pradeep has joined #tripleo03:39
*** pradeep1 has quit IRC03:41
*** pradeep1 has joined #tripleo03:45
*** pradeep has quit IRC03:45
SpamapSstevebaker: ahh the fun of unpublished API's. ;)03:47
*** marun has joined #tripleo03:49
openstackgerritJames Polley proposed a change to openstack/os-cloud-config: If one node is locked, keep trying the rest.
*** pradeep has joined #tripleo03:53
SpamapSanybody ever seen this?03:54
SpamapSMy seed just stops there03:55
*** sdake_ has joined #tripleo03:55
*** pradeep1 has quit IRC03:56
*** nealph_ has joined #tripleo03:58
nealph_lsmola: Ping?03:58
*** tzumainn has quit IRC04:07
*** nealph_ has quit IRC04:15
tchaypoSpamapS: heh04:16
tchaypoSpamapS: haven't seen it, but it's funny because it's happening to someone else.04:16
*** jang1 has joined #tripleo04:17
*** ramishra has quit IRC04:17
SpamapStchaypo: hm, I may have an arch problem04:19
SpamapSsome things i386, some amd6404:20
*** ramishra has joined #tripleo04:20
StevenKNeat. os-cloud-config's highest McCabe complexity is 1304:20
*** ramishra has quit IRC04:20
SpamapShm no04:21
SpamapSStevenK: flirting with "maybe we _COULD_ do this in bash.." ;-)04:22
* SpamapS builds a new seed just to see04:22
StevenKSpamapS: Hah04:23
StevenKSpamapS: Did you see my gentle prod re: mid-cycle?04:23
SpamapSStevenK: no, where did you gently prod?04:29
StevenK[12:41] < StevenK> SpamapS: When you're back, perhaps you should resurrect the Kilo mid-cycle sprint e-mail so we can get some people interested in it04:30
SpamapSStevenK: didn't we delegate that to greghaynes? ;)04:31
SpamapSthis is driving me nuts. Why won't this seed boot?04:31
*** pcrews has quit IRC04:42
*** pradeep1 has joined #tripleo04:45
openstackgerritJames Polley proposed a change to openstack/os-cloud-config: If one node is locked, keep trying the rest.
tchaypoStevenK: through an ept use of patch, I managed to leave in one change from the previous revision that i had intended to revert, and then also add the replacement change that I wanted to use instead04:47
tchaypo^^^ should now be consistent04:47
*** pradeep has quit IRC04:47
*** cwolferh_ has joined #tripleo04:50
*** ramishra has joined #tripleo04:50
* tchaypo attempts an overcloud deployment04:51
tchaypoFOR TEH LOLZ04:51
*** jcoufal has quit IRC04:52
SpamapSok now it's a different error04:55
StevenKSpamapS: That smells like arch mismatch04:56
SpamapSNODE_ARCH=amd64 no_proxy= break=after-error PYPI_MIRROR_URL= SEED_DIB_EXTRA_ARGS="pypi" DIB_NO_PYPI_PIP=1 scripts/devtest_seed.sh04:56
SpamapSafter 'setup-seed-vm -a amd64 -c 2 -m 8192'04:57
StevenKtchaypo: We no longer run nova-bm in CI, so you can't use CI logs to refute04:57
SpamapSand I see amd64 cloud image being pulled in04:57
*** sseago has quit IRC05:01
*** sseago has joined #tripleo05:01
* SpamapS tries clearing .cache/image-create05:08
*** pradeep has joined #tripleo05:10
*** lazy_prince has quit IRC05:11
*** ci-overcloud has joined #tripleo05:11
ci-overcloud************** ci-overcloud complete status=1 ************05:11
*** ci-overcloud has quit IRC05:11
*** pradeep1 has quit IRC05:12
tchaypoERROR: Property error : NovaCompute22: image Error validating value u'': The Image () could not be found.05:13
tchayposo that's new05:13
tchaypoStevenK: I don't understand what nova-bm has to do with anything05:14
*** otoolee has quit IRC05:15
StevenKtchaypo: register_func is called for both05:15
tchaypooh, I see05:17
tchaypois going to return something else05:17
tchaypo*shrug* do I care? I have a problem with how it handles ironic05:17
*** pradeep1 has joined #tripleo05:17
tchaypothis fixes my problem.. nova-bm is deprecated, I'm not sure why I should care enough to fix the problem there05:18
*** pradeep has quit IRC05:19
StevenKtchaypo: I care, which is why I bring it up05:19
*** ramishra has quit IRC05:19
tchaypocurse you.05:25
tchayposo what *does* _update_or_register_bm_node return?05:26
tchaypoi mean, raise?05:26
tchaypoas far as i can tell the answer is "nothing", so this failure is unique to ironic05:27
*** otoolee has joined #tripleo05:27
tchaypoit does:05:27
tchaypo        raise novaexc.ServiceUnavailable()05:27
tchaypobut that's because the *service* is unreachable, not just because *one node* is busy right now. The two are completely different05:27
*** pradeep has joined #tripleo05:28
StevenKtchaypo: It may also raise an uncaught exception from client.baremetal.create()05:28
StevenKBut I'm sort of in conflict. I'd like to remove the nova-bm stuff, but backwards compat :-(05:28
tchayposure, it might do that, I assume it could also raise an uncaught exception from the ironic equivalent, i assume05:29
tchaypobut I05:29
*** pradeep1 has quit IRC05:29
tchaypoI'm not trying to fix the problem of "create failed for unknown reason", I'm trying to fix the problem of "couldn't update right this second because nova has locked the node"05:29
*** pradeep1 has joined #tripleo05:30
tchaypobut that's only because that's the only problem I've seen. If I'd seen other problems maybe I'd care about them too.05:30
tchaypoI'm so sleepy I'm actually contemplating going for a run or something to wake up.05:31
*** ramishra has joined #tripleo05:31
*** pradeep has quit IRC05:32
*** pradeep has joined #tripleo05:36
*** pradeep1 has quit IRC05:37
*** pradeep1 has joined #tripleo05:40
*** pradeep has quit IRC05:42
*** pradeep has joined #tripleo05:43
*** pradeep1 has quit IRC05:45
*** marun has quit IRC05:48
*** pradeep1 has joined #tripleo05:56
*** pradeep has quit IRC05:58
*** pradeep has joined #tripleo05:59
openstackgerritA change was merged to openstack/tripleo-image-elements: Update Horizon to use NEUTRON instead of QUANTUM
openstackgerritSteven Dake proposed a change to stackforge/kolla: Remove nova-base since there is only 1 image for nova-compute
SpamapSnobody has ideas why this VM is doing this?06:03
openstackgerritSteven Dake proposed a change to stackforge/kolla: Make nova-compute work properly with libvirt
*** pradeep1 has quit IRC06:03
SpamapSI'm basically going to assume aliens have replaced my CPU with a frustration generator and leave it at that. :-P06:03
StevenKSpamapS: I had a quick look, but even Google failed me06:05
SpamapSStevenK: yeah me too :-P06:05
StevenKSome randoms have filed bugs with that text on LP, but no replies06:05
lifelessSpamapS: whats going on?06:05
SpamapSlifeless: seed won't boot06:05
StevenKSpamapS: Maybe grep grub2 source?06:05
*** pradeep1 has joined #tripleo06:06
*** dshulyak_ has joined #tripleo06:07
lifelesswhat disk label is in use?06:07
StevenK makes me very very sad06:07
SpamapSdisk label?06:07
*** mrunge has joined #tripleo06:08
SpamapSlifeless: how would I know?06:08
StevenKdisk geometry? No, that isn't right either06:08
*** pradeep has quit IRC06:08
lifelessSpamapS: well, if its out main code, its mbr :)06:09
*** k4n0 has joined #tripleo06:09
SpamapSlifeless: ok. that it is.06:09
lifelesshow much memory is the seed configured with? looks like it might match06:10
uvirtbotDebian bug 605567 in grub-pc "grub-pc: does not boot with "unaligned pointer 0x...." when insmod-ing part_{msdos, gpt}" [Normal,Fixed]06:10
SpamapSI did skip running, because my desire is to setup a seed only.. so I just manually ran setup-seed-vm with the args I wanted06:10
SpamapSlifeless: tried w/ the defaults (3G I think) and 8G06:10
lifelesshmmm, I think we should have mbr, buts its possible folk have been adding gpt06:10
lifelessI haven't tracked closely in a bit06:10
*** pradeep has joined #tripleo06:11
*** pensu has joined #tripleo06:12
SpamapSIt just seems a bit random..06:13
*** pradeep1 has quit IRC06:13
*** pradeep has quit IRC06:15
*** pradeep has joined #tripleo06:16
SpamapSok something weird is going on06:17
lifelessSpamapS: the second one looks like it worked but hung?06:17
SpamapSif I just run' --trash-my-machine on the box, seed comes up06:18
SpamapSso I missed a step06:18
SpamapSok I'll have to chase it down tomorrow06:22
* SpamapS goes to bed06:22
*** pradeep1 has joined #tripleo06:24
*** pradeep has quit IRC06:25
*** marun has joined #tripleo06:26
*** jprovazn has joined #tripleo06:27
sdake_can anyone paste the contents of keystone endpoint-list please?06:29
*** pradeep has joined #tripleo06:30
*** pradeep1 has quit IRC06:31
*** marun has quit IRC06:31
*** pradeep1 has joined #tripleo06:32
*** killer_prince has joined #tripleo06:33
*** marun has joined #tripleo06:33
*** killer_prince is now known as lazy_prince06:33
*** ramishra has quit IRC06:33
*** pradeep has quit IRC06:34
*** pradeep has joined #tripleo06:35
*** avozza is now known as zz_avozza06:35
*** pradeep1 has quit IRC06:36
*** pradeep1 has joined #tripleo06:37
*** ramishra has joined #tripleo06:39
*** pradeep has quit IRC06:40
*** pradeep has joined #tripleo06:40
*** pradeep1 has quit IRC06:41
*** pradeep1 has joined #tripleo06:42
*** pradeep has quit IRC06:44
*** dshulyak_ has quit IRC06:49
*** zz_avozza is now known as avozza06:52
*** dshulyak_ has joined #tripleo06:55
*** pblaho has joined #tripleo07:09
*** pradeep has joined #tripleo07:11
*** pradeep1 has quit IRC07:13
*** proffalken has joined #tripleo07:16
*** proffalken has quit IRC07:16
*** proffalk1n has quit IRC07:17
*** pradeep has quit IRC07:20
*** ifarkas has joined #tripleo07:20
*** pradeep has joined #tripleo07:21
*** gfidente has joined #tripleo07:23
*** gfidente^2nd has joined #tripleo07:23
*** otoolee has quit IRC07:28
*** dshulyak_ has quit IRC07:29
GheRiveromorning all07:29
*** pradeep has quit IRC07:31
*** pradeep has joined #tripleo07:32
*** fabbione has quit IRC07:38
*** ramishra has quit IRC07:41
*** otoolee has joined #tripleo07:42
openstackgerritSteven Dake proposed a change to stackforge/kolla: Properly create endpoints for Nova services
*** gfidente^2nd has quit IRC07:46
*** pradeep1 has joined #tripleo07:49
*** fabbione has joined #tripleo07:50
openstackgerritMotohiro/Yuanying Otsuka proposed a change to openstack/os-cloud-config: Split up extracting Ironic driver_info method
*** jang2 has joined #tripleo07:50
*** pradeep has quit IRC07:52
*** pradeep has joined #tripleo07:58
*** pradeep1 has quit IRC07:58
*** pradeep has quit IRC08:02
*** wuhg has joined #tripleo08:04
shadowerlsmola: ping08:04
*** akrivoka has joined #tripleo08:12
lsmolashadower: pong08:14
shadowerlsmola: so in your testing of the templates with tuskar, did you find any issues with the deployed openstack?08:14
lsmolashadower: so there is instack-test-overcloud script, that I use for testing08:15
lsmolashadower: slagle added there tests of the storage also08:16
lsmolashadower: and seems it is in quite good shape08:16
shadowerlsmola: so the problems you were seeing were mostly because of bad parameters being passed to heat?08:16
lsmolashadower: I will finish the CLI with tuskar today hopefully, so we can easily test also that08:16
shadowerthat'd be cool08:17
openstackgerritGhe Rivero proposed a change to openstack/diskimage-builder: Avoid overwritting of hooks
lsmolashadower: bad parameters, obsolete path for swift08:17
*** sdake_ has quit IRC08:18
lsmolashadower: and the last thing was that you need to increase tiny flavor to 2GB hdd, or the test fails08:18
*** ramishra has joined #tripleo08:18
*** otoolee has quit IRC08:19
gfidentelsmola, shadower btw, looking at the blockstorage/swiftstorage tests08:19
gfidenteshouldn't we run those regardless of SCALE being 0 ?08:19
lsmolashadower: and for the baremetals, we just needed to set correct NTP server and it worked08:19
gfidente*SCALE being > 008:19
*** ramishra_ has joined #tripleo08:19
*** ramishra has quit IRC08:20
lsmolagfidente: hm, I am not sure if it should not fail08:20
lsmolagfidente: does it run storage on the controller node if the scale is 0?08:21
gfidentewell we deploy cinder and swift on controllers when = 008:21
*** fabbione has quit IRC08:22
*** pradeep has joined #tripleo08:22
gfidentegoing to submit a PR for that08:22
lsmolagfidente: ok, I haven't tested it08:22
lsmolagfidente: ok, thank you08:22
gfidenteindeed, we should! :)08:22
*** fabbione has joined #tripleo08:23
*** jp_at_hp has joined #tripleo08:23
*** andrearosa has quit IRC08:23
*** ci-overcloud has joined #tripleo08:23
ci-overcloud************** ci-overcloud complete status=1 ************08:23
*** ci-overcloud has quit IRC08:23
*** jcoufal has joined #tripleo08:24
*** jistr has joined #tripleo08:24
*** derekh has joined #tripleo08:26
*** dmellado_ has joined #tripleo08:26
openstackgerritGonéri Le Bouder proposed a change to openstack/tripleo-image-elements: systemd: always use /lib/systemd
openstackgerritGonéri Le Bouder proposed a change to openstack/diskimage-builder: Revert "introduce $SYSTEMD_SYSTEM_UNIT_DIR"
*** lucasagomes has joined #tripleo08:31
openstackgerritDougal Matthews proposed a change to openstack/tuskar: Put a cap on our cyclomatic complexity
*** fabbione has quit IRC08:32
*** otoolee has joined #tripleo08:33
*** wendar_ has joined #tripleo08:38
*** pradeep has quit IRC08:38
openstackgerritGhe Rivero proposed a change to openstack/tripleo-incubator: Encapsulate getting image for a deployment.
*** wendar has quit IRC08:40
*** fabbione has joined #tripleo08:41
*** ramishra_ has quit IRC08:42
*** ramishra_ has joined #tripleo08:43
*** dtantsur|afk is now known as dtantsur08:47
*** jcoufal has quit IRC08:53
*** jcoufal has joined #tripleo08:54
*** untriaged-bot has joined #tripleo09:00
untriaged-botUntriaged bugs so far:09:00
uvirtbotLaunchpad bug 1381408 in diskimage-builder ""visudo" always got error when building overcloud image" [Undecided,New]09:00
*** untriaged-bot has quit IRC09:00
gfidenteanyone would spare a comment on
gfidentebetween p#5 and p#6 ? later add capability to parse exports file with spaces and in quotes but parsing is not trivial09:01
gfidenteI am not sure if majority prefers to have that or stick with posix portable paths, but I'd like the change to merge one way or another09:01
openstackgerritYuriy Zveryanskyy proposed a change to openstack/diskimage-builder: Add element for build ramdisk with ironic-python-agent
openstackgerritA change was merged to openstack/tripleo-image-elements: sets novnc endpoint uri to https if tls is enabled
*** dmellado_ has quit IRC09:08
*** athomas has joined #tripleo09:13
*** ramishra_ has quit IRC09:14
*** andrearosa has joined #tripleo09:15
*** yamahata has quit IRC09:18
*** ramishra_ has joined #tripleo09:18
*** lazy_prince has quit IRC09:19
*** rlandy has joined #tripleo09:25
*** pradeep has joined #tripleo09:33
*** ramishra_ has quit IRC09:39
*** pelix has joined #tripleo09:44
lxsli+A for please?09:50
*** cody-somerville has joined #tripleo09:50
*** athomas has quit IRC09:58
*** cody-somerville has quit IRC10:02
openstackgerritAlexis Lee proposed a change to openstack/os-apply-config: Extract OacFile class to separate file
*** killer_prince has joined #tripleo10:06
*** killer_prince is now known as lazy_prince10:07
*** athomas has joined #tripleo10:09
*** yuanying_ has quit IRC10:18
*** jcoufal has quit IRC10:18
*** yuanying has joined #tripleo10:19
*** yuanying has quit IRC10:19
*** yuanying has joined #tripleo10:20
*** pradeep has quit IRC10:22
*** yuanying has quit IRC10:24
*** pradeep has joined #tripleo10:32
*** pensu has quit IRC10:45
*** avozza is now known as zz_avozza10:49
*** dmellado_ has joined #tripleo10:50
derekhgreghaynes: you there ? say nothing if your not10:51
derekhgreghaynes: going to push an update to the ci/pip stuff10:51
*** jcoufal has joined #tripleo10:57
*** jcoufal has quit IRC10:57
*** jcoufal has joined #tripleo10:57
*** yuanying has joined #tripleo11:08
*** jerryz has quit IRC11:12
*** pradeep1 has joined #tripleo11:14
*** pradeep has quit IRC11:15
*** derekh has quit IRC11:52
*** ifarkas has quit IRC11:52
*** derekh has joined #tripleo11:56
openstackgerritDerek Higgins proposed a change to openstack-infra/tripleo-ci: Build a local pip mirror of python packages being tested
*** dprince has joined #tripleo12:11
*** mrunge has quit IRC12:11
*** weshay has joined #tripleo12:13
*** rhallisey has joined #tripleo12:17
gfidentelxsli, does look okay btw or can you spot any problem there?12:20
*** pensu has joined #tripleo12:20
*** pensu has quit IRC12:25
*** jdob has joined #tripleo12:31
lxsligfidente: looking12:38
*** andreaf has quit IRC12:38
*** andreaf has joined #tripleo12:38
lxslip sure you'll need an ExtraConfig at least :)12:41
*** nosnos has quit IRC12:41
*** nosnos has joined #tripleo12:42
*** pensu has joined #tripleo12:43
*** nosnos has quit IRC12:46
*** zz_avozza is now known as avozza12:54
*** k4n0 has quit IRC12:55
*** tzumainn has joined #tripleo12:55
greghaynesderekh: Ah, that variable isnt documented12:58
greghaynesbut if it works then fine with me :)12:58
openstackgerritAlexis Lee proposed a change to openstack/tripleo-heat-templates: Add block completion signal
openstackgerritRyan Brady proposed a change to openstack/tripleo-heat-templates: Add ExtraConfig param to storage templates
openstackgerritAlexis Lee proposed a change to openstack/tripleo-heat-templates: Add block completion signal
*** ramishra has joined #tripleo12:59
lxsligfidente: please :)13:00
gfidentelxsli, yeah I can't +313:00
gfidenteI would though13:00
gfidentegreghaynes, when I asked the other guys why this change was needed in nova/controller13:08
gfidentethey told me it was obvious13:08
gfidenteit'd be nice if I could pick that up again13:08
gfidentebut I am with you13:09
greghayneswell, we do still need a blurb, maybe that one was in error13:09
gfidenteit is not obiouus13:09
gfidentegreghaynes, see my comment on p#2413:10
derekhgreghaynes: that latest patch is testing a ZUUL_CHANGES=openstack/python-keystoneclient:master:refs/changes/99/99999/913:11
*** morazi has joined #tripleo13:11
derekhgreghaynes: I'll double check the results to make sure everything happened as needed13:11
gfidenteso purpose of change is passing specific values for a config key to some roles13:11
greghaynesgfidente: Yea, I would have agreed with you but I always like more explanation ;)13:12
greghaynesderekh: Oh? I was trying to figure out a sane way to test that code13:12
*** akrivoka has quit IRC13:13
openstackgerritRyan Hallisey proposed a change to stackforge/kolla: Docker image and process cleaup tool
gfidentegreghaynes, I'll a sentence in commit message indeed, so I can take care of lxsli comment too13:13
gfidentethanks :(13:13
gfidenteoh sorry that was a :)13:13
derekhgreghaynes: I've just put ZUUL_CHANGES=openstack/python-keystoneclient:master:refs/changes/99/99999/9 in the patch, which of course mean another revision is needed to take it out again, but its the only way I could think of testing it13:14
greghaynesyep, good idea13:14
*** pensu has quit IRC13:17
openstackgerritgreghaynes proposed a change to openstack/tripleo-heat-templates: Add SSL PKI properties for keystone
*** openstackgerrit has quit IRC13:19
*** openstackgerrit has joined #tripleo13:19
openstackgerritGiulio Fidente proposed a change to openstack/tripleo-heat-templates: Passthrough{Specific} and allNodesConfig for BlockStorage nodes
rhalliseysdake, morning13:23
sdakereviews in the queue ;)13:25
sdakealmost get all of nova working13:25
sdakeone last little bug to sort out13:25
openstackgerritGiulio Fidente proposed a change to openstack/tripleo-heat-templates: Passthrough{Specific} and allNodesConfig for BlockStorage nodes
rhalliseyexcellent :)13:25
sdakemaybe you can help13:26
sdakeI was fairly stumped last night13:26
sdakebut it was bedtime so that had somethign to do with it13:26
rhalliseywhat are you running into?13:26
sdakesomething about an endpoint not found13:26
sdakeyet the endpoints are set properly in the db13:26
sdakeI had a dream maybe it was not a nova endpoint but something else13:27
sdakeso I'll investigaet that today13:27
rhalliseysdake, so I have keystone mariadb and nova-api running13:28
sdakeyou need glance as well13:28
sdakenova contacts othe rendpoints tho maybe it requires a cinder endpoint registration13:28
sdakeand you need all the changes in the review queue as well13:31
*** ramishra has quit IRC13:32
sdakemarket on fire today13:33
sdakege up 3.5 %13:33
sdakecrazy bump for ge13:33
*** marun has quit IRC13:34
openstackgerritTzu-Mainn Chen proposed a change to openstack/tuskar-ui: Add script to generate node csv
derekhslagle: jp_at_hp do ye know if anybody has ever looked at getting CI on the icehouse branch working ?^openstack/tripleo.*+branch:stable/icehouse,n,z13:40
derekhif not should we?13:40
*** noslzzp has joined #tripleo13:41
* derekh can take a look if we care13:41
*** ramishra has joined #tripleo13:42
sdakerhallisey larsks here is what I get from nova-api:13:43
sdakethis tells me nova-api is configured improperly13:43
larskssdake: Let me take a look.13:44
sdakethis on a nova list op13:44
jp_at_hpderekh: I made a very few attempts to get a build to run on icehouse and failed miserably.  I haven't had any time to dig in either...13:44
larskssdake: Hold on, that's from nova list?13:45
sdake(in nova-api)13:46
sdakenova list prints out auth failed13:46
slaglederekh: the last time i looked, it wasn't using the stable/icehouse branches of the tripleo projects13:46
slaglederekh: that's what was causing all the problems13:46
sdakebut to reproduce you need the review queue changes, because ot herwise the endpoints are busted13:46
derekhslagle: yup, I have an idea to make it work if we still care I can give it a go13:46
slaglederekh: the devstack-gate job was checking them out though, not sure why the right thing wasn't used13:47
*** cody-somerville has joined #tripleo13:47
slaglederekh: if it's easy to try, sure. but i wouldn't spend a bunch of time on it13:47
slaglederekh: i didn't even end up create stable branches for juno13:47
larskssdake: pulling images, please hold :)13:48
slaglederekh: given the new backwards compat approved spec, i'd like to see us setup jobs that use tripleo master to deploy stable/juno of evertying else13:48
derekhslagle: ok, I'll do that instead, unless anybody else is gonna13:48
sdakelarsks oh, and glance doens't seem to actualy store cloud images for me13:48
larsksHuh, works for me. At least, it writes files into /var/lib/glance/images, and they show up in glance list.13:49
derekhso any thoughts on those icehouse patches stuck. should we just approve them or abandon////13:49
slaglederekh: i wasn't planning on it in the short term. i added a note about this discuss at the summit. but, it seemd no one had a problem with it when we were talking about what jobs to add at the midcycle13:49
slagleso if you wanted to take a go at it, i wouldn't complain :)13:50
sdakeusing glanceclient, no love13:50
sdakeyou just wrot ethem directl yto /var/lib/glance/images?13:50
larskssdake: No, glance does that.13:50
sdakeI was like "huh" :)13:51
larskssdake: Run "glance image-show 17456857-fa48-4bfe-b859-0df89fa71758"13:51
larsksand tell me what you see.13:51
sdakesee log above, that is command I'm using13:51
larsksAlso, is it actually creating a file in /var/lib/glance/images? Or not?13:51
larsksIf yes, try adding --is-public=True to your image-create command and see if it shows up in "image-list"13:52
jp_at_hpderekh: does it make sense to abandon existing patches and see what the plan of running master against icehouse service branches yields?13:52
sdakeright just a sedf13:52
sdakeforgot about t hat option13:52
larsksSo it *looks* like it is actually working.13:52
larsksMy experience with recent versions of glance is that without --is-public=True images do not show up in the list.13:52
larsksAt least, not my default.13:52
larsksI haven't investigated that much yet, because on my own system I don't care if everything is public or not.13:52
sdakelarsks yup that fixed it (is-public=true)13:53
slaglejp_at_hp: i'm not opposed to abandoning those patches. it would certainly be interersting to see if tripleo master could still deploy icehouse13:53
sdakeok there is one problem solved13:53
sdakesee that only took you 3 minutes :)13:53
sdakeI struggled for hours last night on that annoyance13:53
derekhjp_at_hp: slagle ok, how about we abandon them and I'll aim to get a job running to run tripleo-* master with juno of openstack releases (and I'll try icehouse as an experiment)13:54
slaglederekh: wfm. will abandon today13:55
openstackgerritNicholas Randon proposed a change to openstack/tripleo-heat-templates: Simplify the layout of haproxy net binds config
SpamapSmy seed built w/o rabbitmq?14:05
greghaynesSpamapS: It needs to be defined in SEED_DIB_EXTRA_ARGS IIRC14:06
SpamapSgreghaynes: ... why?14:06
greghaynesalso, do you really *need* rmq? :p14:06
greghaynesthere was a reason, I forget14:07
SpamapSnova-api has to talk to nova-conductor _SOMEHOW :)14:07
SpamapSOk so yeah I used that variable to add pypi to use my mirror14:07
SpamapShow silly of me14:07
SpamapSgreghaynes: probably to allow building w/ qpid ;)14:07
SpamapSFailed ot spot that. Doh14:08
greghaynesthings like this make me wish we used something like envdir14:09
SpamapSgreghaynes: doooo it14:09
greghaynesIts not a real project until you pull in some djb tools14:09
greghaynesthen everyone runs away screaming14:09
*** rlandy has quit IRC14:11
*** bnemec is now known as beekneemech14:12
*** ramishra has quit IRC14:12
openstackgerritA change was merged to stackforge/kolla: Remove nova-base since there is only 1 image for nova-compute
lxslineeds a mailreader14:12
openstackgerritA change was merged to stackforge/kolla: Make nova-compute work properly with libvirt
*** radez_g0n3 is now known as radez14:13
*** wuhg has quit IRC14:13
openstackgerritA change was merged to stackforge/kolla: Properly create endpoints for Nova services
sdakeonlyh two more components to debug plus jpeeler's work :)14:15
sdake\o/ we are close folks :)14:15
sdakejpeeler I recommend looking a tthe design pattern in glance or nova-controller for heat-api & heat-engine14:15
jpeeleryeah i've redone it quite a bit14:16
sdakespecifically the use of the base image14:16
openstackgerritLars Kellogg-Stedman proposed a change to stackforge/kolla: fix nova auth configuration
sdakelarsks does that fix api auth?14:18
larskssdake: Probably.  Building images right now.14:19
sdakeare you building to kollaglue?14:20
jpeeleri was wondering about that14:20
sdakeif not, might as well14:20
jpeeleri copied devstack's config, which uses 35357 instead of 500014:21
larsksjpeeler: I am pretty sure that auth does not require access to the admin api.14:21
sdakeit does not14:21
larsksjpeeler: But I will let you know in a few minutes if it worked or not.14:21
jpeelerok i'll just wait then14:21
sdakelarsks are you building nad pushing to kollaglue?14:22
larskssdake: I am not.  For the most part, I think it's best if pushes to kollaglue are only done by one person, and I am fine if that person is you.14:22
sdakeI'll go ahead and pull down that change and build it14:22
larsksIt would be nice if the images in the kollaglue namespace are only ever build from stackforge master (e.g., wait for the change to merge before pushing into kollaglue/ namespace)14:23
sdakeagree - I think after our rush on m1, we can set some more sane parameters (to be discussed at monday meeting) - i'll add to agenda14:23
sdakeset 10-15 minutes to discuss policy on pushing/etc14:24
sdakedentist gave me norco 10/325614:24
sdakenormally he is really tight with medications I'm not sure what came over him14:25
larsksIf we can have someone install webhooks on the stackforge repositories, we can set up our own automatic build environment while we wait for openstack-infra.14:25
larsksAnd then stop doing manual pushes to kollaglue/.14:25
sdakelarsks infra doesn't rely on github, it relies on its own repos and then pushes to github later14:26
sdakethey would like to remove github entirely, but I think the likelyhood of that being accepted by the community is low :)14:26
larskssdake: That doesn't really impact what I just said at all :).  As long as the pushes get to github in a reasonable amount of time, *we* could rely on it pending the necessary support in infra.14:26
larsksOr we can have something set up that just polls perodically if we can't get automated notification.14:27
sdakeoh, I thought you meant some automatic github plugin or something14:27
sdakeya poll is easier I think14:27
clarkbwe wont set up webhooks in github. you can however use a gerrit account to run tests based on gerrit events14:27
*** jdob1 has joined #tripleo14:27
clarkbthis is a well known and tested process with tooling to go along with it14:27
larsksclarkb: docs pointer, please?14:28
clarkbalso what are you waiting on us for?14:28
larsksclarkb: mordred had said something about setting up infrastructure for supporting automated docker builds.  If we can do this ourselves that is awesome.14:28
clarkbWe give you root on test slaves. not sure what else is missing14:29
clarkbwe have trusty and f20 nodes. both should be able to docker14:29
larsksclarkb: that sounds sufficient. How do we go about getting that set up?14:29
clarkbsubmit changes to openstack-infra/project-config to add jobs under jenkins/ and tell zuul to run them in zuul/layout.yaml14:30
*** otoolee has quit IRC14:30
clarkbjayf and jroll may have ideas too since they are dockering in a job of theirs14:31
*** jehb has joined #tripleo14:31
*** rwsu has joined #tripleo14:31
larsksclarkb: if I am looking to run something explicitly *after* a review has been approved/merged...can I still do that using that mechanism?14:31
sdakeclarkb we aren't waiting on you, we are waiting on us :)14:31
*** spzala has joined #tripleo14:31
clarkblarsks yes that is what jayf and jroll do14:31
sdakelarsks I intend to tackle it in milestone #214:32
*** morganfainberg has quit IRC14:32
larsksclarkb: Okay, thanks, I will ping them.14:32
sdakeor if you want, you can as well14:32
clarkbsdake if you arent waiting on anything dont bother with third party ci14:32
larsksYeah, I don't think that's what we want.14:32
sdakeI dont think we were intending to use third party ci ?14:32
sdakewe want to use infra ci14:32
sdakewe just haven't got the gates I want implemented yet because of time constraints14:32
clarkbsdake github hooks is basically third party ci14:33
sdakeyes, I think larsks wants github hooks, I dont want that14:33
larskssdake: that is not what I said.14:33
sdakeso no reason to argue :)14:33
larskssdake: if we can do this in -infra, then we are all set.14:33
sdakeit doesn't matter, we are all in agreement about what needs to be done14:33
openstackgerritRyan Brady proposed a change to openstack/tripleo-heat-templates: Adding SNMP related parameters to storage templates
openstackgerritRyan Brady proposed a change to openstack/tripleo-heat-templates: Add converted version of block and object storage
openstackgerritRyan Brady proposed a change to openstack/tripleo-heat-templates: Compute and controller templates without
*** morganfainberg has joined #tripleo14:33
sdakeanyone been watching agents of shield14:34
sdakethat show rocks14:34
sdakelarsks I think before we automatically push the images we want a gate to make sure what we push is not busted14:35
sdakejpeeler and I have been talking about  this for milestone 2 or perhaps milestone 314:35
sdakewe need it soon14:35
larsksI think we need to stop manual build/push as soon as possible. I think we also need the CI on those images, but I would not make that a blocker.14:36
sdakeits a littile more complicated though, because there is a race between upstream's rdo packaging changes14:36
sdakeI agree we need ci asap14:36
sdakethis manual testing is terrible14:36
sdakeit  took us 4 months to get ci going in heat btw14:36
sdakeand I was actively trying to get it implemented across the project from month 214:37
sdakeci is just hard to bootstrap, not sure why14:37
larskssdake: speaking of manual testing, with my rebuilt images I can successfully "nova list".14:37
larsksSo that seems to work now.14:37
sdaketime to boot a vm :)14:37
larsksThe / script I've been dropping into these images is really useful for testing.14:37
sdakehow do you use it to test btw?14:38
sdakesome nsenter magic?14:38
sdakemight consider a tools/* tool to help run it14:38
larsksI run "docker-enter <container-id> /", where docker-enter is a wrapper around nsenter (
sdakeI'll try that out once I rebuild14:39
larsksNo output == it worked.14:39
larsksYou get errors otherwise.14:39
sdakelarsks in heat, jpeeler actually implemented our first test cases :)14:40
sdakeprior to that, we did all manual testing for ~4mo14:40
sdakeI prefer not to repeat that14:40
sdakealthough manual testing for this is much easier14:40
sdakeeither it works or it does not work14:40
sdakewith heat, we had test plans etc before we finally got the 2.5k test cases implmeented14:41
larsksWell, we have three test cases now (for keystone, nova, and glance), so guess we're ahead of the game.14:41
larsksWe just need some sort of harness to run things :)14:42
sdakeyup that is going to be a real challenge I think14:42
sdakeI have tried to think a way to do it, but its just going tohav e to be a hack14:42
sdakeit will be as difficult as writing all of the code we have written thus far14:42
sdake(which admittidely is not that complex)14:43
sdakejpeeler can you ack larsks change - it seems to work for him14:43
openstackgerritA change was merged to stackforge/kolla: fix nova auth configuration
*** pensu has joined #tripleo14:43
* jpeeler copies that change14:44
greghayneswell, thats annoying
greghayneswe dont really have a lot of info for debugging when the demo vm fails14:44
openstackgerritRyan Brady proposed a change to openstack/tripleo-heat-templates: Add ExtraConfig param to storage templates
larsksjpeeler: you're working on heat, right?14:45
jpeelerlarsks: yes14:45
larsksDo you know if heat actually requires both auth_uri *and* auth_{host,port,protocol}?14:45
larsksOr will it work with just auth_uri?14:45
larsksI am curious if it behaves the same as Nova.14:45
jpeelerno i don't, devstack has all of them though14:46
larsksI guess I can just test it locally.14:46
* larsks goes to edit heat.conf14:46
jpeelerlet me know what happens14:46
greghayneslarsks: you mean identity_uri?14:46
larsksgreghaynes: I dunno.  The documentation is littered with auth_uri, but I have seen the other as well.14:47
greghaynesbecause AIUI identity_uri superceded auth_{host,port,protocol}14:47
larsksgreghaynes: Spiffy.14:47
larsksAny idea about the purpose of auth_uri vs identity_uri?14:47
openstackgerritBen Nemec proposed a change to openstack/tripleo-image-elements: Remove vif_plugging workaround
*** jdob1 has quit IRC14:49
greghayneslarsks: no idea about auth_uri, I think thats going to be app specific14:49
greghaynesidentity_uri is a keystone config thing14:49
larsksjpeeler: well, heat appears to need the various auth_* components.  But, using the public api (port 5000) works fine.14:49
*** otoolee has joined #tripleo14:50
greghaynesso the auth_{host,port,protocol} is deprected, not a huge deal but if youre making new config I would just use identity_uri in place of them14:51
*** akrivoka has joined #tripleo14:52
*** jp_at_hp has quit IRC14:56
*** pcrews has joined #tripleo14:57
openstackgerritJeff Peeler proposed a change to stackforge/kolla: (Still WIP at the moment, but much closer)
larsksgreghaynes: thanks.14:58
openstackgerritgreghaynes proposed a change to openstack/tripleo-image-elements: Use keystone identity_uri instead of auth_*
*** untriaged-bot has joined #tripleo15:00
untriaged-botUntriaged bugs so far:15:00
uvirtbotLaunchpad bug 1382468 in tuskar "Bad parameter exception" [Undecided,New]15:00
*** untriaged-bot has quit IRC15:00
uvirtbotLaunchpad bug 1381408 in diskimage-builder ""visudo" always got error when building overcloud image" [Undecided,New]15:00
*** pbelanyi has quit IRC15:08
*** dtantsur has quit IRC15:08
openstackgerritGiulio Fidente proposed a change to openstack/tripleo-heat-templates: Passthrough{Specific} and allNodesConfig for BlockStorage nodes
gfidentelxsli, I just fixed the dependencied there, thanks for noticing :(15:17
*** yamahata has joined #tripleo15:18
*** akrivoka has quit IRC15:26
*** pradk has joined #tripleo15:28
*** otoolee has quit IRC15:29
jpeelersdake: do you know what i should look for if docker logs doesn't return anything? i should at least be getting an error15:29
larsksjpeeler: try running it by hand with "docker run" and provide all the necessary environment variables manually?15:30
larsksjpeeler: Can I pull the images and try them locally?15:30
sdakechmod +x start.sh15:30
larsksOr what sdake said :)15:31
sdakeor any other scripts you have15:31
jpeelerargh, but yes larsks they are pushed15:31
sdakeanother option when you run into that, is you can run it locally eg docker run kollaglue/fedora-rdo-heat-api15:31
sdakeitwill show you precisely what is failing15:31
sdakeI recommend trying that right now just for a learning experience15:31
sdakeyou will llikely see " permission denied"15:32
mordredsdake, larsks: I have the plan in my head of what we need to do to get the docker building/publishing done, I just haven't done it15:32
mordredsdake, larsks: so what I will try to do instead is write it down so that if you want to do it instead of waiting for me, you can15:32
sdakewiki page would rock mordred15:33
larsksmordred: cool.  Let me know if you need a hand with anything.  I am wanting to tackle that Real Soon Now :)15:33
sdakewould be happy to coalesce around one solution15:33
mordredthe main issue that needs to be solved is that we need to split being able to run docker build on one machine and docker push on a different one15:33
larsksmordred: why that?15:33
larsksbuild machine does not have outbound access?15:33
clarkbmordred: but we have already solved that for all of the other artifacts we publish15:33
clarkblarsks: it won't have credentials15:34
mordredclarkb: right. we have to solve it for docker15:34
larsksclarkb: got it.15:34
mordredclarkb: the pattern is likely the same15:34
jpeelersdake: yep, thanks15:34
larsksSo...push to private repo on another machine, and from there to docker hub?15:34
mordredlarsks: the pattern we do elsewhere15:34
mordredis that we publish the artifact to tarballs.o.o - and then have a job on a trusted machien that knows how to pull it from tarballs.o.o and push it to the appropriate publication place15:35
mordredthis is slightly more complex with docker15:35
mordredbecause the artifact is most efficient in docker15:35
mordred(we could export/import, but that seems very lame)15:35
*** sirushti has left #tripleo15:36
larsks...because export/import would collapse all the image layers, right?15:36
mordredso it might be that we want to run a dockerhub that we trust about as much as tarballs.o.o so that pushing to it from a host is not a thing we're worried about15:36
mordredand then have a job that pulls from that and pushes to the public dockerhub15:36
larsksThat sounds reasonable to me.15:36
mordredin any case - I'll write up a thing with more words15:37
derekhgreghaynes: looks like I choose the wrong project to test with because keystoneclient is installed from the git repo already, going to redo it with os-apply-config15:37
mordredI don't think it's likely to be _hard_ ... just a little bit of work that needs to be done15:37
*** otoolee has joined #tripleo15:37
openstackgerritDerek Higgins proposed a change to openstack-infra/tripleo-ci: Build a local pip mirror of python packages being tested
clarkbmordred: oh I hadn't considered the layering aspect (I was thinking just push image up, grab then publish. the ironic agent image job does it this way)15:38
greghaynesderekh: haha, so that means it worked, right? ;)15:42
openstackgerritAlexis Lee proposed a change to openstack/os-apply-config: Extract OacFile class to separate file
derekhgreghaynes: well it didn't not work ;-)15:43
openstackgerritAlexis Lee proposed a change to openstack/os-apply-config: Set owner+group from control file
*** funzo has quit IRC15:43
*** funzo has joined #tripleo15:44
openstackgerritRamakrishnan G proposed a change to openstack/diskimage-builder: iso element to build bootable ISO images
openstackgerritJames Slagle proposed a change to openstack/diskimage-builder: svc-map requires PyYAML
sdakelarsks nova list working  for me15:47
sdake\o/ :)15:47
sdakenow for the million dollar operation15:47
openstackgerritAlexis Lee proposed a change to openstack/os-apply-config: Set owner+group from control file
SpamapSmordred: why don't you write up the dib-docker change as a spec?15:52
openstackgerritNicholas Randon proposed a change to openstack/tripleo-heat-templates: Simplify the layout of haproxy net binds config
*** bth- has joined #tripleo15:56
jlabocki1bth - daneyon has been working on neutron docker/kubernetes configurations as well. You two should talk.15:56
mordredSpamapS: that would bea  great idea15:57
bth-jlabocki1, sounds good15:57
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Add package install support for tuskar
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Add package install support for tuskar-ui
openstackgerritNicholas Randon proposed a change to openstack/tripleo-heat-templates: Simplify the layout of haproxy net binds config
*** derekh has quit IRC16:09
*** marun has joined #tripleo16:16
*** eghobo has joined #tripleo16:16
*** dmellado_ has quit IRC16:18
*** sdake_ has joined #tripleo16:19
jpeelermy image is based on fedora-rdo-base, so how am i getting /opt/kolla/ No such file or directory?16:24
*** pblaho has quit IRC16:29
*** eghobo has quit IRC16:29
rhalliseyjpeeler, you need add kolla-common to the image16:33
jpeelerit should be in kollaglue/fedora-rdo-base, right?16:34
rhalliseyjpeeler, take a look at nova-ctr-base/Dockerfile16:34
rhalliseyADD config-nova-controller.sh16:34
jpeelersdake: is not +x either16:35
jpeelerrhallisey: notice how sources kolla-common.sh16:36
larsksjpeeler: kolla-common is not supposed to be +x.16:36
larsksIt is sourced, not executed.16:37
jpeelerok, so how's what i'm experiencing possible?16:37
larsksI don't know what you're experiencing...just sat down after grabbing lunch :)16:37
larsksjpeeler: I wonder if you are pulling an outdaded fedora-rdo-base?16:38
larsksI don't know how recently the kollaglue/ one was rebuilt.16:38
jpeeleri'm assuming everything in the kollaglue namespace is current, should i just rebuild it?16:38
larsksProbably. Because we don't have an automated build process, there's no guarantee that what's in kollaglue is current.16:39
jpeelerhere it goes16:39
jpeeler(the building)16:40
*** lucasagomes is now known as lucas-dinner16:42
*** pelix has quit IRC16:43
sdake_larsks any idea on
sdake_jpeeler rebuild your fedora-rdo-base16:44
sdake_i had the same problem16:44
sdake_or do docker pull kollaglue/fedora-rdo-base16:44
larsks sdake_ What exactly do you want me to look at there?  It's failing to connect to neutron. Do you have neutron running?16:45
sdake_shoudl actuallly get it done16:45
sdake_larsks well i'm curious if you have a suggestion as to what we should be doing to get the network rolling16:45
larskssdake_: k8s does a pull before launching...suggesting that maybe we need to push a new fedora-rdo-base to docker hub.16:45
sdake_this is caused by a nova boot16:45
larskssdake_: you need somehting providing a network api, either neutron or nova network.16:45
sdake_larsks if jpeeler is building his images - it will use his fedora-rdo-base cache16:45
larsksRight now the configuration file requires neutron.16:45
larskssdake_: True dat.16:45
sdake_any idea how to configure nova network16:46
sdake_since neutron isn't ready16:46
larsksHaven't touched nova-network for a couple years now; I just use neutron.16:46
larsksSo I would have to read the docs.16:46
sdake_well my network fu isn't great16:46
sdake_if you could tackle that, I'd appreciate it :)16:46
larsksI will take a look.16:46
sdake_I htink we a re close16:46
sdake_larsks do you think nova networking would work for this scneario?16:48
sdake_radez any help you can provide here would be appreciated as well16:49
larsksI'm not entirely sure. I think the networking aspect is going to be one of the trickiest.16:49
sdake_i'd be happy if w e could jus tboot a vm, and not necessarily be able to use it16:49
sdake_(eg if the network didn't work, that would be ok in the short term)16:50
larsksThat is, I think we can probably make things "work" enough to boot a server, but not necessarily to give it proper connectivity.16:50
larsksSo good, we are on the same page there :)16:50
sdake_ya that wfm16:50
sdake_we can fix that in m216:50
sdake_assuming daneyon can get neutron rolling we will jus tuse that in m216:50
sdake_and I have high confidence given his current work so far16:50
sdake_extra gold star for networked vm :)16:51
*** athomas has quit IRC16:51
*** pensu has quit IRC16:52
*** akrivoka has joined #tripleo16:56
*** akrivoka has quit IRC16:57
*** akrivoka has joined #tripleo16:57
*** akrivoka has quit IRC16:57
*** jprovazn has quit IRC16:59
*** dkehn_ is now known as dkehn17:01
*** sdake_ has quit IRC17:01
*** pensu has joined #tripleo17:02
*** rameshg87 has joined #tripleo17:02
rameshg87SpamapS: hi17:04
daneyonit may be easier to get nova-net to work than neutron l3/dhcp/ovs.17:04
SpamapSrameshg87: hello. :)17:04
SpamapSdaneyon: don't do it17:04
rameshg87SpamapS: would require your review again as i had rebase because of merge conflict17:04
SpamapSdaneyon: that zombie will never die if you keep feeding it17:04
rameshg87SpamapS: please have a look at it :)17:04
SpamapSrameshg87: will do.17:05
daneyonfrom what i see, we need to extend a separate network to the minions for either nova-net or neutron to work.17:05
rameshg87SpamapS: thanks ....17:05
daneyonthat's what i'm working on right now.17:05
rameshg87dprince: hi17:05
sdakelarsks daneyon here is my devstack flatdhcp network config - which parts are the important settings?
sdakedaneyon we agreed a bit above to just get a booting vm would be a good first step17:06
sdakemakign it have connectivity a good wsecond step17:06
daneyonSpamaps: I hear ya. I go back and forth with it. I agree with you on one hand. On the other hand, many current prod deployments use nova-net and we should support their req's. I think people will let nova-net dies when neutron provides enough stability, scalability, etc..17:07
daneyonsdake: 3-6, 33-3517:10
sdakethanks I'll try those17:10
daneyonsdake: it's been a while since i've touched nova-net. give me a few min to review some of my older configs.17:10
*** dshulyak_ has joined #tripleo17:11
SpamapSdaneyon: we've taken a strong stance here in the deployment program that we deploy the defaults and we fix the defaults if they're broken. Neutron is the default now.. so..17:11
daneyonSpamapS: Understandable.17:11
SpamapSdaneyon: no problem with providing nova-net as an option, of course.17:11
SpamapSBut if we keep adding peripheral projects that only support nova-net and not neutron .. nova-net will never die.17:12
rameshg87tripleo-cores, can someone else also please have a look at . this is required for an ironic driver released in juno.17:12
daneyonSpamaps: I'm working on neutron, so I definitely want to see it be successful. If things go well with kolla, I wouldn't be surprised if we get pressure to support nova-net until the community in general begins embracing neutron.17:13
daneyonSpamapS: OK, I think we're on the same page.17:14
SpamapSYeah that's the right thing, I think. Let people deploy it with Neutron, and when they go "oh noes I need nova-net" do that. :)17:14
SpamapSBut really, the last big thing, multi-host, is dead now.17:14
SpamapSNeutron has DVR now which does all that multi-host did17:14
daneyonSpamaps: What about us though? Neutron will not be ready until milestone-2 (hopefully), so are we going to try using nova-net in the meantime?17:15
larsksdaneyon: That's what sdake  and I were discussing...he just wants to get a nova server booting, so he needs *something* to provide the network api.17:15
daneyonlarsks: that's what i was thinking. just want to make sure we're all on the same page.17:16
rameshg87SpamapS: thanks a lot for quick review. :-) ; please let me know if i can ping someone else too for a review.. we need this badly for the ironic driver released for juno.17:17
SpamapSdaneyon: Neutron won't be ready, wha?17:17
*** jogo is now known as flashgordon17:17
SpamapSwhats broken in Neutron, or is it just that the requirements for getting ovs working in docker is hard?17:18
daneyonreq's to get neutron working in a container. i cut neutron-server earlier this week. i'm working on the necessary changes to heat-kube template to support multiple interfaces on minions and containers, then build the neutron l3/dhcp/ovs containers.17:19
daneyonwe moved the neutron bp to milestone 2 the other week because of the LOE.17:20
*** pradeep1 has quit IRC17:22
sdakelarsks I just booted a vm17:23
larsksWoo woo17:23
sdakebut nova list doesn't appear to work (based upon daneyon's lines I  copied)17:23
sdakehouston we have liftoff -> ebbf084f-7f7a-4db0-b2ce-6653b60c0bd017:24
*** dshulyak_ has quit IRC17:24
sdakedaneyon ideally we want this to work on bare metal not just heat-kubernetes template17:25
larskssdake: that does'nt look like "liftoff".17:25
larsksThat looksl like "failed to lift off"... (status == ERROR)17:26
daneyonsdake: for sure. i just figured we need it to work in heat-kube since that's the current dev solution of choice.17:26
larsksdaneyon: the what now?  I did not follow where heat-kube came up...17:27
*** pcrews has quit IRC17:27
*** morganfainberg is now known as captainmorgan17:29
rhalliseysdake, that's awesome17:30
daneyonlarsks: i was thinking that neutron l3 agent containers have a 2nd interface to support floating-ips and public traffic. The 2nd interface would get mapped to a kub-ext bridge on the minion. The kube-ext bridge gets mapped to a 2nd interface on the minion. The 2nd interface on the minion maps to a public-facing network. Thoughts?17:30
larsksdaneyon: question: where does that second interface come from? does kubernetes have a mechanism for setting that up?  I haven't looked into that in any detail yet.17:31
larsksdaneyon: yooooooo hoooooo...17:34
*** pradeep has joined #tripleo17:34
openstackgerritRyan Brady proposed a change to openstack/tripleo-heat-templates: Add ExtraConfig param to storage templates
*** pradeep has quit IRC17:35
SpamapSdaneyon: they don't have to have a second interface. They just have to have addresses to support the provider network where the floating ips live.17:36
*** pradeep has joined #tripleo17:36
daneyonlarsks: nova boots the kube fedora image with 2 nic's. nic 1 is for managing the kube host and for providing transport between the vxlan endpoints (just like what we do today). The 2nd nic will get mapped to the new kube-ext bridge on the minion. kube-ext will be a flat bridge that provide connectivity to the neutron l3 agent containers and bridge traffic from the l3 container through the kube host eth1 to the external container17:37
daneyon network. Do you see any issue with this approach?17:37
sdakelarsks any idea on this problem:
sdakethis is sudo docker ps keystone17:38
sdakekeystone spins for like 3-4 minutes wiating for itself17:38
sdakeit would be nice if we could speed that up17:38
larsksdaneyon: I am still unclear how the neutron l3 containers are connected to the kube-ext bridge.17:38
larskssdake: I don't think so.  We start keystone and then just wait for it to become active so that we can register endpoints.17:39
larsksThe solution, of to move one time configuration out of the container :)17:39
larsksBecause then our would be a lot simpler.17:40
larsksBut it might not actually make things any faster.17:40
larsks(That is, if it takes keystone a while to become active here, it would probably still take keystone a while to become active...)17:40
SpamapSIs keystone spinning while it creates its tables?17:41
SpamapSThat always takes longer than I think it should.17:41
larsksSpamapS: I figured the tables would be created by the "db_sync" operation.17:41
SpamapSthats kind of what I was asking.. but poorly ;)17:41
larsksWhich happens before we start up keystone and start waiting for it.17:41
daneyonlarsks: have a 2nd interface on the container that gets connected to kube-ext?17:41
larsksSo by the time you see the "waiting for keystone" message, db sync was already run.17:41
larsksdaneyon: Which brings me back to my original question: where does that second interface come from?17:42
*** rameshg87 has left #tripleo17:43
*** pradeep1 has joined #tripleo17:44
openstackgerritLars Kellogg-Stedman proposed a change to stackforge/kolla: remove glance-pod.json
*** dshulyak_ has joined #tripleo17:45
*** pradeep has quit IRC17:46
sdakelarsks my q is why does it spin for 3-4 minutes?17:46
sdakeit doesn't take that long for keystone to become active does it ?  I think there is a bug in the checking code possibly17:46
larskssdake: I'm not sure.  I don't think I've seen that behavior.17:47
larsksI don't think there's a bug in the checking code...I mean, it's very simple.17:47
sdakecheck out fpaste above17:47
larsksYes, I looked at it.17:47
daneyonlarsks: create an image with 2 NICs. eth1 would have the standard ovs bridge config (no ip addr). Take an approach similar to pipework or netcfg to connect eth1 to kube-ex?17:47
sdakedidn't it look suspect?17:47
larskssdake: well, there are no timestamps, but it only sleeps for 1 second each, so that looks like about 10 seconds.  No?17:48
larsksdaneyon: I apologize, but I am terribly  confused.  Kubernetes is responsible for setting up the networking inside a container.  How do you propose to create a container with multiple interfaces? Can kubernetes do this?17:48
larsksOr is that not what you meant by "have a 2nd interface on the container"...?17:49
sdakeoh 1 sec17:49
sdakeok then17:49
sdakeI thought it was 10 seconds17:50
sdakeits just if I do keystone endpoint-list it takes forever to get rolling17:50
sdakeeven though keystone is running17:50
sdakeI'm one of those performance maniac guys :)17:50
larsksI would not worry about performance right now.17:50
sdakeI jsut dont want it to be so slow that its unusable17:51
sdakewhich isn't the case17:51
*** yamahata has quit IRC17:51
*** eghobo has joined #tripleo17:52
daneyonlarsks: i'm still grasping kube, so hang in there with me. If kube is unable to create multiple interfaces, then could we have the l3-agent use a single nic. Have this nic connect to kube-ext. config the nic as the bridge interface and create a bridge within the container with a IP for mgt. something like this:
larsksdaneyon: I think that right now we have very little flexibility in terms of pod networking.17:53
larsksI think you get one interface, and it gets connected to the same bridge as all the other interfaces.17:54
larsksI think the way we solve this for now is by creating bridge devices inside the neutron container that are *not* bridged to actual NICs.17:54
*** pradeep has joined #tripleo17:54
larsksAnd then we use NAT rules for outbound connectivity.17:54
larsksAnd we worry about making inbound work after we have things booting and with functional outbound.17:54
*** pradeep1 has quit IRC17:55
SpamapSHow does it do the ovs networking method?17:57
SpamapSI see that it replaces docker0 with obr017:57
larsksSpamapS: That is bascially just a name change.17:57
larsksAnd the overlay network is something set up by the heat templates (e.g., not set up by k8s itself).17:58
daneyonlarsks: with that said, could we create a bridge within the l3 container that does vxlan overlay over the standard kube bridged network to a separate device/node that acts as a vxlan gateway?17:58
larsksdaneyon: Yeah, sure, maaaaaaybe :).  I had that thought to.  But I would first focus on what I said a moment ago...just get things running with outbound working via 'iptables -t nat -A POSTROUTING -j MASQUERADE'17:59
SpamapSlarsks: yeah I see that now..
daneyonlarsks: are you aware of any kube dev to extend the current networking model?18:00
larsksdaneyon: there is an open issue regarding external connectivity, but that is a different problem, I think.18:00
larsksI am not aware of any work that would cover the neutron use case, necessarily.18:01
larsks...but I haven't looked or asked, either.18:01
sdakelarsks ignore noise about keystone performance18:02
larskssdake: No worries.18:02
sdakeI tested, it takes about 1 min to boot from start, which is n't terrible18:02
sdakeI thought it was like  5minutes18:02
sdakewhich is less desireable18:02
openstackgerritRyan Hallisey proposed a change to stackforge/kolla: Docker image and process cleaup tool
daneyonlarsks: OK, i figured i would ask before digging.18:02
*** radez is now known as radez_g0n318:04
SpamapSSeems like the problem is kubernetes is assuming you will only ever care about L3 in pods.18:06
SpamapSlooking at the code, you probably just have to add an option to add arbitrary bridges.18:07
*** radez_g0n3 is now known as radez18:07
greghaynesSpamapS: So, trying to brainstorm a process for how to move helion specific code out into a helion role18:08
SpamapSgreghaynes: that might not be the right approach18:08
SpamapSgreghaynes: I think the way to go is to move mechanics into pods18:08
larsksSpamapS: kubernetes is heavily application-centric, and in that model, a single interface is probably fine.  It might require some convincing to get upstream to accept greater functionality.  Worth a shot, though.18:08
SpamapSgreghaynes: and then Helion specific mechanics have their own pods. But Helion specific changes to the mechanics.. are forks.18:09
SpamapSlarsks: Yeah, or as daneyon says, just tunnel out to something on each host.18:09
sdakelarsks are you certain pods share the network namespace?18:09
greghaynesnot sure what you mean by mechanics, got an example?18:09
larskssdake: Yes, I am totally sure.18:10
sdakeok so I'm a bit jammed up18:10
larskssdake: to be clear: containers *in a pod* share a network namespace.18:10
sdakewith this iptables thing18:10
larsksdifferent pods have different network namespaces.18:10
SpamapSoh now that's interesting.18:10
larsksSpamapS: The shared network namespace for containers in a pod?18:11
greghaynesSpamapS: also, I think theres some term interference happening as pod is a docker thing, right?18:11
SpamapSbut ultimately, you need to get a whole range of floating IPs routed into each network node. That seems like you would need collaboration from the host.18:11
SpamapSlarsks: yeah I didn't know that.18:11
larsksSpamapS: It is a surprisingly nice abstraction -- because it makes it easier to follow a "one process per container" sort of model, even for things that are reasonably tightly linked (because "localhost" is the pod, not the container).18:12
SpamapSgreghaynes: sorry yeah we're going to have to multiplex w/ the docker stuff. :)18:12
SpamapSlarsks: totally18:12
larskssdake: What is the current iptables issue?18:12
openstackgerritSteven Dake proposed a change to stackforge/kolla: Change from using Neutron to Nova flat networking
SpamapSgreghaynes: I said pod, but I meant roles.18:12
sdakelarsks paste coming hold a sec18:13
SpamapSI wonder if you shouldn't just plan to run the L3 agent on the host.18:13
openstackgerritpatrick-crews proposed a change to openstack/tripleo-image-elements: Add missing packages to tempest element
larsksSpamapS: Or possibly inside a non-kubernetes-managed container, to avoid polluting the host filesystem.18:14
larskssdake: We are also going to need to add a "nova-network" image to run the nova-network service.18:15
SpamapSlarsks: yeah .. though really, it seems like there is an argument to be made for kubernetes to support "l2-capable-pods" .. so have a pod just for l3 agent and it can have its normal netowrk in the same overlay as the non-l2-capable pods.18:15
larsksSpamapS: ...and slowly but surely kubernetes begins to transform into OpenStack :)18:16
sdakelarsks nova-compute cycles with this error:
SpamapSbut without all that python baggage. ;)18:16
sdakeinstead we ge t go baggage - yay !18:17
sdakeour atomic image is about 400mb18:17
greghaynesSpamapS: oh, huh, seems I totally misunderstood roles18:17
sdakethis is the full os for running k8s + docker18:17
larskssdake: You maybe need to install openstack-nova-network also in order to get the necessary python modules?18:17
sdakeI would expect an import error?18:17
larsksThat *is* an Importerror.18:18
sdakeI'll give ita try tho18:18
larsksLine 13.18:18
sdakefast and loose today - always like this on release day :)18:18
greghaynesSpamapS: So we would make a role for galera, for example18:18
sdakelarsks duh thanks18:18
SpamapSgreghaynes: more like, we would make a role for 'galera-single-node-down' and 'galera-single-node-up' and 'galera-cluster-shutdown' and 'galera-cluster-recovery'18:20
SpamapSgreghaynes: we make roles for desired states.18:20
greghaynesReally? Seems like theres a pattern of passing in a desired state to a role18:20
SpamapSgreghaynes: and then we take the servers through said states .. and that's an update. :)18:21
SpamapSgreghaynes: we _could_ do that18:21
SpamapSI just don't love it.18:21
SpamapSbut I may not have fully embraced the encapsulation.18:21
greghaynesI am just reading random roles on ansible-galaxy ATM18:21
SpamapSgreghaynes: we can't really use those for much inspiration18:21
SpamapSgreghaynes: they're focused on turning an empty box into a "foo"18:22
SpamapSgreghaynes: we're focused on turning a "foo" into a "foo.1" :)18:22
SpamapSthough eventually I'd love for us to also be able to do initialization18:22
greghaynesHrm, ok. So then for the Helion vs upstream split18:23
greghaynesHelion specific bits would be in a helion-gelera-cluster-recovery role maybe?18:24
SpamapSHelion should never have specific galera bits.18:24
SpamapSwe abused controllerMgmt for that.18:24
greghaynesYep ;)18:24
SpamapSHelion will likely run extra services though.18:24
greghaynesok, that can be variable-ified though18:24
SpamapSMy intention is to move all of that away from ansible18:24
SpamapSand into the images18:24
SpamapSBut short term yeah, just have forked variables18:25
*** pcrews has joined #tripleo18:25
greghaynesSweet, ill play around with roles, see what sticks18:25
SpamapSgreghaynes: cool18:26
*** pradeep1 has joined #tripleo18:30
*** pradeep has quit IRC18:33
openstackgerritJames Slagle proposed a change to openstack/diskimage-builder: svc-map requires PyYAML
*** pradeep1 has quit IRC18:36
*** cody-somerville_ has joined #tripleo18:40
*** cody-somerville_ has joined #tripleo18:40
*** cody-somerville has quit IRC18:41
*** dshulyak_ has quit IRC18:44
sdakelarsks nova-network communicates with nova how, amqp?18:47
larsksYes.  nova-{scheduler,conductor,api,network} all use the rpc message bus.18:48
sdakeshouldn't network be part of compute, or you think controller?18:48
larsksI think nova-network should be a separate container, just like nova-{scheduler, conductor}, etc.18:48
sdakeyes I got that part18:48
larsksOh, which pod. I see.18:48
larsksUhhhhh.  I would make it part of the controller right now.18:48
sdakeI think we will end with compute-neutron and compute-nova18:49
sdakecompute-novanet or something18:49
larsksAnd think you are probably right.18:49
larsksAnd actually, we can make it part of the compute pod, turn on multi-host, and see how that works.18:49
sdakeI think I'll probably put in compute pod for now18:49
sdakewhat is multi-host, you mean minon count greater then 1?18:49
larsksNova network has two modes of operation.18:49
larsksYou can have a single instance, or you can have one nova-network service per compute host.18:50
larsksIf you make nova-network part of the compute pod, I think you're going to need to use multi-host mode.18:50
sdakegot it18:50
sdakeok I'll crank that out after a smoke break :)18:50
sdake(building images now anyway, can't do much)18:50
*** sdake_ has joined #tripleo18:55
*** pensu has quit IRC18:56
daneyonsdake: the other day you said you were having issues with a newer release of kube? what version and what was the issue? the reason i ask is because i am having issues after rebuilding my kube cluster. all the services are running and look good but no minions in my list.19:01
larsksdaneyon: note that new, incompatible kube packages are dropping and/or have dropped today.  If you have 0.4, all bets are off.19:02
daneyonlarsks: looks like i'm still using 0.3 -> 0.3-0.619:03
larsksOkay, that should be okay.19:03
openstackgerritJeff Peeler proposed a change to stackforge/kolla: (Still WIP at the moment, but much closer)
*** pensu has joined #tripleo19:12
*** radez is now known as radez_g0n319:17
sdake_daneyon that particular problem requires yo uto restart kube-apiserver and kube-scheduler19:17
sdake_sudo systemctl restart kube-apiserver19:17
radez_g0n3sdake: I almost suggested trying nova network in our meeting monday19:17
*** radez_g0n3 is now known as radez19:18
radezI don't have much experience with it though19:18
sdake_radez do you know if nova net is in controller or compute in rhos?19:18
radezI don't know for sure.19:18
radezit's been so long19:19
sdake_anyone you can ask?19:19
sdake_I don't know the right person to query on the subject19:19
radezdyokum would be a good resource or some of the nova guys19:19
sdake_and I don't recall from my foreman root canal the details19:19
sdake_i jus tmean as structured in foreman19:19
radezsdake_:  there's an doc on it19:21
sdake_daneyon the version that has the problem is whatever is installed after doing a  heat create19:21
larskssdake_: nova-network goes with nova-compute + you enable multi-host.19:22
larsksThat is a True Fact.19:22
sdake_what about neutron?19:22
sdake_will we end up with neutron server with each compute?19:22
radezjust the plugin agent on each compute host for neutron19:22
sdake_ok sounds like we have a plan then19:23
radezis that different for ml2?19:23
larsksSo, with Juno neutron now has a distributed virtual router (DVR) feature, with which I am not familiar.  Pre-Juno, neutron is like nova-network when *not* in multi-host mode -- you have one network host, and all traffic flows through there.19:23
sdake_i'll add a new container19:23
openstackgerritLars Kellogg-Stedman proposed a change to stackforge/kolla: introduce a common base for compute and controller
larskssdake_: take a look at that before you add a new container -- I was trying to facter out common stuff between nova-controller and nova-compute.19:24
daneyonsdake: looks like i can see my minions. now i need to figure out what i changed to solve the problem. i have restart all the master/node services multiple times with no luck.19:26
larsksdaneyon: is working now, but unclear why?19:26
daneyonlarsks: correct. i noticed etcd was running on the minions. i think etcd should only be running on the master, so I stopped the service and restarted kube services. don't know why that would be an issue since minions = /etc/kubernetes/config:KUBE_ETCD_SERVERS=http://$MASTER_IP:400119:29
*** jcoufal has quit IRC19:29
larsksdaneyon: That's correct, etcd should only run on the master.19:29
larsksI wonder if there is a racy startup between etcd and the k8s services or something.19:29
larsksIf we're lucky, that particular issue will simply disappear with 0.4 (and I'm sure we'll have several new issues, too).19:30
*** cody-somerville_ has quit IRC19:31
*** pcrews has quit IRC19:31
daneyonlarsks: well that's not the issue anyways. i have 1 minion running etcd and 1 not. both are configured /etc/kubernetes/config:KUBE_ETCD_SERVERS=http://$MASTER_IP:4001 and both minions appears in the master minion list. maybe theirs some service dependencies on the master. I was restarting all services and it seemed like things worked when i only restarted api and scheduler.19:32
sdake_larsks looks good - radez or jpeeler can one of you review larsks last review request19:32
openstackgerritA change was merged to stackforge/kolla: remove glance-pod.json
openstackgerritA change was merged to stackforge/kolla: introduce a common base for compute and controller
openstackgerritA change was merged to stackforge/kolla: Docker image and process cleaup tool
openstackgerritA change was merged to stackforge/kolla: Adds Neutron Server Support
openstackgerritLars Kellogg-Stedman proposed a change to stackforge/kolla: fix script name in nova-controller images
sdake_jpeeler your heat change looking really good19:49
sdake_I submitted a few comments19:49
openstackgerritRyan Hallisey proposed a change to stackforge/kolla: Add nova-novncproxy to the nova-controller
*** yuanying has quit IRC19:56
openstackgerritA change was merged to stackforge/kolla: fix script name in nova-controller images
sdake_what does novncproxy do anyway19:57
larsksit proxies vnc connections from a browser to the appropriate compute host19:58
sdake_i see19:58
larsksThat is how the instance console works.19:58
sdake_that explains why there is x graphics in the dep chains i guess19:58
sdake_never used instance cnsole19:58
sdake_use virt-manager for that19:58
larsksYou can't always use virt-manager if you're interacting with a remote openstack deployment of which you are not the admin :)19:59
sdake_wfm, not for a real deployment19:59
rhalliseylarsks, should I add it to the controller?19:59
rhalliseylarsks, I figured I would19:59
larsksrhallisey: well, we will need an image for it somewhere at some point, so it can't hurt.20:00
larsksI think maybe we will ultimately want to run it wherever we run horizon.20:00
larsksBut I'm not sure.20:00
openstackgerritRyan Hallisey proposed a change to stackforge/kolla: Add nova-novncproxy as a kube service
*** dprince has quit IRC20:06
*** sdake_ has quit IRC20:08
*** sdake_ has joined #tripleo20:08
sdake_i feel like the tab police20:09
*** pcrews has joined #tripleo20:11
sdake_do we need both port and containerPort in service files?20:11
*** rhallisey has quit IRC20:12
larskssdake_: Yes.  When is the "inside port" on the container to which the proxy connects, the other is the port on which the proxy listens.20:13
larskssdake_: Fix the bashate gate and you will not have to be the tab police :)20:13
larskss/When/One/ that was weird20:13
openstackgerritA change was merged to stackforge/kolla: Add nova-novncproxy as a kube service
sdake_jpeeler ignore my comment then about the port not being necessary :)20:16
sdake_pretty sure that change was busted above20:18
*** jdob has quit IRC20:20
openstackgerritSteven Dake proposed a change to stackforge/kolla: Change to using flat nova networking from neutron
sdake_hey guys the repo has 255 commits20:21
sdake_2^8 ftw !20:21
sdake_new build workflow rocks20:24
*** pensu has quit IRC20:26
*** rhallisey has joined #tripleo20:28
*** jehb has quit IRC20:30
-openstackstatus- NOTICE: Gerrit will be offline from 2100-2130 for project renames20:35
*** ChanServ changes topic to "Gerrit will be offline from 2100-2130 for project renames"20:35
*** cwolfe__ has joined #tripleo20:36
*** cwolferh_ has quit IRC20:40
sdakelarsks what might haautooproxy do20:43
*** rhallisey has quit IRC20:47
*** pweiss has joined #tripleo20:47
openstackgerritPradeep Kilambi proposed a change to stackforge/kolla: WIP: Add Ceilometer service support
SpamapSmordred: flamel trick:  select horribly formatted json in vim, type "!flamel -|flamel -" ... reformats nicely :)20:55
SpamapSI should probably just add a --reformat or something20:55
*** rwsu has quit IRC20:56
SpamapSactually "jq ." kinda does that better, but doesn't preserve order. :)20:56
*** jistr has quit IRC20:58
*** untriaged-bot has joined #tripleo21:00
untriaged-botUntriaged bugs so far:21:00
uvirtbotLaunchpad bug 1382468 in tuskar "Bad parameter exception" [Undecided,Triaged]21:00
uvirtbotLaunchpad bug 1381408 in diskimage-builder ""visudo" always got error when building overcloud image" [Undecided,New]21:00
*** untriaged-bot has quit IRC21:00
*** radez is now known as radez_g0n321:01
sdake503 on reviews ftl21:04
-openstackstatus- NOTICE: Gerrit is offline from 2100-2130 for project renames21:04
*** ChanServ changes topic to "Gerrit is offline from 2100-2130 for project renames"21:04
* sdake groans21:04
sdake30 minute break time21:04
*** weshay has quit IRC21:07
clarkbsdake: use the gertty :P21:13
*** sdake_ has quit IRC21:14
*** sdake_ has joined #tripleo21:16
sdakeprefer breaktime clarkb :)21:19
*** sdake_ has quit IRC21:20
*** ChanServ changes topic to "CI: Presumed O-K | Using OpenStack to deploy OpenStack; meetings Tuesday 1900/0800 UTC in #openstack-meeting-alt"21:27
-openstackstatus- NOTICE: Gerrit is back online21:27
*** captainmorgan is now known as needscoffee21:29
*** needscoffee is now known as captainmorgan21:29
openstackgerritJonathan Brownell proposed a change to openstack/os-apply-config: Treat null values in JSON as being absent rather than 'None'
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Enable dracut deploy ramdisks
*** bth- has quit IRC21:38
*** andreaf has quit IRC21:59
*** gfidente has quit IRC22:01
*** openstackgerrit has quit IRC22:03
*** openstackgerrit has joined #tripleo22:04
*** jang2 has quit IRC22:26
*** gfidente has joined #tripleo22:30
*** andreaf has joined #tripleo22:41
*** rwsu has joined #tripleo22:52
openstackgerritGiulio Fidente proposed a change to openstack/tripleo-image-elements: Improve nfs-server to support multiple shares and options
*** marun has quit IRC23:09
openstackgerritDaneyon Hansen proposed a change to stackforge/kolla: Fixes k8s neutron pod image source and formatting

Generated by 2.14.0 by Marius Gedminas - find it at!