Wednesday, 2016-03-02

*** ShannonM has quit IRC00:01
openstackgerritKevin Carter proposed openstack/openstack-ansible-os_nova: Implement base configuration for independent repository  https://review.openstack.org/28673600:07
-cloudnull- if you have review cycles the following reviews could use your left over cycles https://review.openstack.org/#/q/status:open+branch:master+topic:bp/independent-role-repositories and https://review.openstack.org/#/q/status:open+branch:liberty+topic:liberty_upgrade -- TYVM :)00:09
stevelleand the gate plays Sad Trombone for us00:21
*** jmckind has joined #openstack-ansible00:23
*** skamithi has joined #openstack-ansible00:28
jmccroryhmmm so --no-download is only in 14.x virtulenv00:30
*** thorst_ has joined #openstack-ansible00:30
jmccroryan option in*00:31
cloudnullyes ... its never-download00:33
stevelledevilish details, yeah00:33
openstackgerritKevin Carter proposed openstack/openstack-ansible-repo_build: Updated venv build process for virtualenv compat  https://review.openstack.org/28692300:35
cloudnullthat issue is caused by the updated constraints in the repo-server limiting us to virtualenv from global requirements.00:36
cloudnulljmccrory stevelle ^^ that should fix it for us00:36
stevellecloudnull: watching it, but it's happy hour :)00:37
*** evilrob has left #openstack-ansible00:37
cloudnullits passing local checks.00:37
* cloudnull babysitting the gate.00:37
*** thorst_ has quit IRC00:38
*** thorst_ has joined #openstack-ansible00:38
lane_konghi, guys, i am working on developing ansible role for zaqar service,  I know a little ansible, but not very familiar with. I came across a problem, please see http://paste.openstack.org/show/488879/00:39
lane_kongthe project structure of the ansible role is very like other services.00:40
cloudnulllane_kong:  it would seem that your venv path is not being detected as a venv00:40
*** thorst_ has quit IRC00:40
*** thorst_ has joined #openstack-ansible00:41
*** help has joined #openstack-ansible00:41
lane_kongcloudnull: thanks for reply. but how it could be a venv?00:41
*** help is now known as Guest5279200:41
*** jwitko has quit IRC00:41
lane_kongthere is just a directory call 'bin' in  /openstack/venvs/zaqar-untagged00:41
lane_kongcloudnull: this is the zaqar_install.yml: http://paste.openstack.org/show/488881/00:43
lane_kongcloudnull: appreciate if you could give some advise00:43
cloudnulllane_kong:  so the venv directory doesnt have all of the other bits ?00:46
cloudnulllane_kong:  do you have a role i can pull in and try out ?00:46
lane_kongcloudnull: i forked from https://github.com/stevelle/openstack-ansible-zaqar, made a few mofication internally00:48
lane_kongwe want to deploy zaqar using ansbile in our comany00:48
lane_kongi am afraid the original repo didn't work00:49
lane_kongcloudnull: if you need , i can paste all the file of the role for you00:49
*** phalmos has quit IRC00:50
cloudnulllane_kong: can you provide a diff from the chanage you have locally00:50
cloudnulli can patch the original role and see if i can find something00:50
lane_kongcloudnull: yes, sure, wait for a moment00:50
*** scarlisle has joined #openstack-ansible00:58
*** woodard has quit IRC01:00
*** jguy_ has joined #openstack-ansible01:01
*** woodard has joined #openstack-ansible01:01
*** sdake has joined #openstack-ansible01:03
*** sdake has quit IRC01:16
*** scarlisle has quit IRC01:17
*** sdake has joined #openstack-ansible01:31
cloudnullanyone around to shove this through https://review.openstack.org/#/c/286923/01:32
cloudnullstevelle jmccrory ^01:32
jmccrorygood to go01:33
*** sdake has quit IRC01:33
openstackgerritMerged openstack/openstack-ansible-repo_build: Updated venv build process for virtualenv compat  https://review.openstack.org/28692301:33
cloudnullso once thats through we can move on to the next problem :)01:34
cloudnullwhatever that is ...01:34
openstackgerritKevin Carter proposed openstack/openstack-ansible: Implements IRR for the repo_build role  https://review.openstack.org/28489601:34
stevellelet's roll the dice and see01:35
*** izaakk has quit IRC01:37
*** Mudpuppy has joined #openstack-ansible01:38
lane_kongcloudnull: appreciate your help anyway01:41
lane_kongyou helped me find what's the problem is01:42
cloudnullanytime01:42
*** fawadkhaliq has quit IRC01:42
stevellechecking back in a bit on that gate. got a jam session scheduled01:47
lane_kongcloudnull: one more question, is there any guidance for creating pre-build venv?01:53
*** thorst_ has quit IRC01:56
*** thorst_ has joined #openstack-ansible01:57
*** scarlisle has joined #openstack-ansible02:01
*** thorst_ has quit IRC02:05
cloudnulllane_kong: so with your changes + what ive done here https://github.com/os-cloud/openstack-ansible-os_zaqar02:07
cloudnullthe role should work02:07
cloudnullto  build the venv in an isolated fashion set -e zaqar_developer_mode=true02:07
cloudnullthat will build the zaqar source and install it in a local venv02:08
cloudnullwithout any of the additional OSA machinery02:08
lane_kongcloudnull: man, thanks for much!02:08
lane_kongso much02:08
lane_kongcloudnull: i will test it, hope it could work for me02:09
*** Guest52792 has quit IRC02:12
*** weezS has quit IRC02:13
*** furlongm_ has quit IRC02:21
*** furlongm has joined #openstack-ansible02:32
*** jmckind is now known as jmckind_02:40
*** daneyon has joined #openstack-ansible02:42
*** daneyon_ has quit IRC02:45
*** jmckind_ has quit IRC02:46
*** daneyon has quit IRC02:47
*** woodard has quit IRC03:04
*** thorst_ has joined #openstack-ansible03:04
*** thorst_ has quit IRC03:10
*** sdake has joined #openstack-ansible03:15
*** weshay has quit IRC03:22
*** scarlisle has quit IRC03:22
cloudnullif any cores are around can we bang this though https://review.openstack.org/#/c/284896/ <- stevelle andymccr odyssey4me d34dh0r53 jmccrory automagically mattt hughsaunders03:22
*** jmckind has joined #openstack-ansible03:22
cloudnullwhich will finally unblock the gate.03:23
* cloudnull going to eat will check back later. maybe03:23
stevelle^ done03:24
*** jmckind is now known as jmckind_03:29
prometheanfirearen't you a helper03:30
*** sigmavirus24 is now known as sigmavirus24_awa03:34
*** jmckind_ has quit IRC03:35
*** sdake has quit IRC03:39
*** jmckind has joined #openstack-ansible03:39
*** fawadkhaliq has joined #openstack-ansible03:43
*** sdake has joined #openstack-ansible03:53
openstackgerritJimmy McCrory proposed openstack/openstack-ansible-repo_server: Use slurp to collect the repo_server ssh keys  https://review.openstack.org/28604203:54
cooljjmccrory: myself and others ran into the same issue even with the slurp patch https://bugs.launchpad.net/openstack-ansible/+bug/153178304:02
openstackLaunchpad bug 1531783 in openstack-ansible liberty "os_nova not using all configured compute_hosts" [High,Triaged] - Assigned to Jesse Pretorius (jesse-pretorius)04:02
jmccrorycoolj, yeah. it seems to happen if you have any host in the group involved that fails before that task04:03
*** cemmason has joined #openstack-ansible04:04
jmccroryand trying to use a limit breaks it as well04:04
*** shausy has joined #openstack-ansible04:06
*** thorst_ has joined #openstack-ansible04:08
*** jmckind has quit IRC04:09
jmccrorycoolj do you still have a deployment that's failing because of that?04:10
*** skamithi has quit IRC04:11
cooljjmccrory: i was helping the deployment team troubleshoot, and i don't recall who we were adding nodes for. i can follow up with them tomorrow and see if we have any ongoing or that have hit it recently04:14
*** thorst_ has quit IRC04:16
*** gfa is now known as gfa_04:18
jmccroryoh ok, also ran into it a team here a week or two ago. thinking adding a '| default("")' to the end of this line will help, will test it tomorrow https://github.com/openstack/openstack-ansible-os_nova/blob/master/tasks/nova_compute_key_distribute.yml#L1904:18
openstackgerritJimmy McCrory proposed openstack/openstack-ansible-repo_server: Removed sshd_config file  https://review.openstack.org/28441004:20
*** Mudpuppy has quit IRC04:39
*** jmckind has joined #openstack-ansible04:39
*** Mudpuppy has joined #openstack-ansible04:40
*** furlongm has quit IRC04:45
*** furlongm has joined #openstack-ansible04:48
openstackgerritMerged openstack/openstack-ansible-repo_server: Removed sshd_config file  https://review.openstack.org/28441004:50
openstackgerritMerged openstack/openstack-ansible-repo_server: Use slurp to collect the repo_server ssh keys  https://review.openstack.org/28604204:56
*** fawadkhaliq has quit IRC04:58
*** cemmason has quit IRC04:59
*** furlongm has quit IRC05:02
*** sdake has quit IRC05:09
*** furlongm has joined #openstack-ansible05:09
*** thorst_ has joined #openstack-ansible05:14
*** thorst_ has quit IRC05:21
*** furlongm has quit IRC05:25
*** cemmason has joined #openstack-ansible05:31
*** admin0 has joined #openstack-ansible05:33
*** pcaruana has joined #openstack-ansible05:36
*** admin0 has quit IRC05:37
*** javeriak has joined #openstack-ansible05:37
*** admin0 has joined #openstack-ansible05:37
*** furlongm has joined #openstack-ansible05:40
*** pcaruana has quit IRC05:49
*** cemmason has quit IRC05:54
*** cemmason has joined #openstack-ansible05:54
*** Mudpuppy has quit IRC05:56
*** jmckind has quit IRC05:59
*** javeriak has quit IRC06:07
*** furlongm has quit IRC06:13
*** thorst_ has joined #openstack-ansible06:19
*** furlongm has joined #openstack-ansible06:21
*** phiche has joined #openstack-ansible06:25
*** thorst_ has quit IRC06:26
*** furlongm has quit IRC06:26
*** furlongm has joined #openstack-ansible06:29
*** furlongm has quit IRC06:34
*** gfa_ is now known as gfa06:37
*** jamielennox is now known as jamielennox|away06:38
*** furlongm has joined #openstack-ansible06:38
*** furlongm has quit IRC06:39
*** furlongm has joined #openstack-ansible06:43
*** admin0 has quit IRC06:49
*** jamielennox|away is now known as jamielennox06:49
*** phiche has quit IRC06:57
*** furlongm has quit IRC06:58
*** furlongm has joined #openstack-ansible06:58
*** phiche has joined #openstack-ansible07:04
*** jamielennox is now known as jamielennox|away07:08
*** furlongm_ has joined #openstack-ansible07:16
*** furlongm has quit IRC07:18
*** thorst_ has joined #openstack-ansible07:23
openstackgerritMatt Thompson proposed openstack/openstack-ansible-os_aodh: Implement base configuration for independent repository  https://review.openstack.org/28665507:28
openstackgerritMatt Thompson proposed openstack/openstack-ansible-os_glance: Implement base configuration for independent repository  https://review.openstack.org/28666707:29
openstackgerritMatt Thompson proposed openstack/openstack-ansible-os_heat: Implement base configuration for independent repository  https://review.openstack.org/28669307:30
*** thorst_ has quit IRC07:31
openstackgerritMatt Thompson proposed openstack/openstack-ansible-pip_install: Test version of pip being installed  https://review.openstack.org/28528407:32
*** mikelk has joined #openstack-ansible07:47
*** markvoelker has quit IRC07:58
*** admin0 has joined #openstack-ansible08:00
*** pcaruana has joined #openstack-ansible08:02
*** raddaoui has joined #openstack-ansible08:15
*** jamielennox|away is now known as jamielennox08:25
*** thorst_ has joined #openstack-ansible08:29
*** raddaoui has quit IRC08:30
*** thorst_ has quit IRC08:35
*** permalac has joined #openstack-ansible08:42
*** permalac has quit IRC08:42
*** winggundamth has joined #openstack-ansible08:42
*** zhangjn has quit IRC08:43
openstackgerritMerged openstack/openstack-ansible-os_nova: Implement base configuration for independent repository  https://review.openstack.org/28673608:46
*** zhangjn has joined #openstack-ansible08:48
*** markvoelker has joined #openstack-ansible08:58
*** permalac has joined #openstack-ansible09:00
*** javeriak has joined #openstack-ansible09:01
*** asettle has joined #openstack-ansible09:05
winggundamthI'm using stable/liberty and have problem install repo oslo right now09:06
winggundamth  Could not find a version that satisfies the requirement oslo.i18n<1.6.0,>=1.5.0 (from ceilometer) (from versions: 3.1.0)09:06
winggundamthNo matching distribution found for oslo.i18n<1.6.0,>=1.5.0 (from ceilometer)09:06
winggundamthwhat is the workaround for now?09:06
winggundamthshould I change to use master instead? I see a lot of major change version so I think it can break the installation09:07
openstackgerritMerged openstack/openstack-ansible-os_cinder: Implement base configuration for independent repository  https://review.openstack.org/28668209:17
matttwinggundamth: i'd not recommend using master09:17
matttwinggundamth: fine for development, not for a live deployment09:17
matttwinggundamth: it sounds like we may have some dependency issues, let me build a liberty installation now and see if i can troubleshoot09:17
matttwinggundamth: are you on a specific liberty tag?09:17
odyssey4memattt based on some of what gus was saying last night, I have the feeling that the pip/wheel/setuptools trio is causing breakage and I think we need to get the appropriate pinning back into liberty / kilo as currently we're pip<8 but wheel and setuptools are not properly pinned to ensure compatibility09:20
odyssey4meeffectively wheel is building wheels that pip can't install because of the change in names - same as kilo (except we pinned wheel in kilo)09:20
mattt:(09:20
matttmaster is moving again ?09:20
admin0guys, no change was done in the current branch(liberty) right :D?09:21
odyssey4memattt not sure yet - waiting for some rechecks to confirm, but it looks like the necessary changes have merged09:24
odyssey4meadmin0 changes happen from time to time - you can see it with https://review.openstack.org/#/q/status:merged+project:openstack/openstack-ansible+branch:liberty or simply in the git history09:25
winggundamthmattt: yes. I'm using liberty09:25
odyssey4meadmin0 do you have a specific question there?09:25
admin0question was .. all the gate/changes of yesterday, where do they end up ?09:25
matttodyssey4me: any idea why https://review.openstack.org/#/c/286694/ needed to fix a flake8 violation?  why was that not being picked up before?09:25
mattt^^^ jmccrory if you're awake :)09:25
matttwinggundamth: great -- but are you using a specific tag?09:26
winggundamthmattt: I'm doing git clone -b liberty https://github.com/openstack/openstack-ansible /opt/openstack-ansible. is this answer that you want?09:27
matttwinggundamth: yeah that helps, thanks :)  i'll do a build now to test09:27
odyssey4memattt I guess our choices are to ignore E501 violations, to implement a longer max line length as in https://review.openstack.org/#/c/286696/4/tox.ini or to not do '#!{{ swift_venv_enabled | bool | ternary(swift_venv_bin + "/", "/usr/bin/env ") }}python' at the top of some of the python files?09:28
odyssey4meor, of course, to just ignore particular files - I'mnot really in favour of this option as it leads down a dark path09:28
winggundamthmattt: http://eavesdrop.openstack.org/irclogs/%23openstack-ansible/%23openstack-ansible.2015-07-23.log.html I'm seeing that someone said the rackspace repo missing that oslo version09:29
*** thorst_ has joined #openstack-ansible09:33
odyssey4mewinggundamth the rpc-repo doesn't have to have it - it comes from pypi if that repo doesn't have it09:33
*** cemmason1 has joined #openstack-ansible09:34
*** cemmason1 has quit IRC09:34
winggundamthodyssey4me: ic so I'm trying to dig into that right now09:34
*** cemmason1 has joined #openstack-ansible09:34
winggundamthI think packages version 1.5-1.6 has been missing. https://pypi.python.org/pypi/oslo.i18n09:35
winggundamthhttps://pypi.python.org/pypi/oslo.i18n/1.5.0 it still there09:36
openstackgerritMerged openstack/openstack-ansible-os_neutron: Implement base configuration for independent repository  https://review.openstack.org/28669409:36
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_horizon: [WIP] Implement base configuration for independent repository  https://review.openstack.org/28665009:36
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_horizon: Implement base configuration for independent repository  https://review.openstack.org/28665009:37
*** cemmason has quit IRC09:37
*** thorst_ has quit IRC09:40
odyssey4memattt what do you think of implementing https://review.openstack.org/286591 in all the roles? it gives us the equivalent of verbose output, but in a readable format09:42
matttodyssey4me: yeah seems logical09:44
winggundamthmattt: this is error I got from ansible in case you want more detail http://paste.openstack.org/show/488918/09:44
odyssey4mewinggundamth an option you can go with is to checkout to the last tag in the liberty branch - if the head of the branch is broken then we can fix that up (we have a known issue to deal with), but the last tag should hopefully still work09:45
*** lane_kong is now known as lynn09:45
*** lynn is now known as Guest2997009:46
*** Guest29970 is now known as lynn_kong09:46
matttwinggundamth: cool, running my build now09:46
winggundamthodyssey4me: but from my git clone -b liberty command. it should checkout the last tag of liberty branch right?09:49
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone: [WIP] Implement base configuration for independent repository  https://review.openstack.org/28666009:51
odyssey4mewinggundamth the best thing to do for now is to wait for mattt's test to confirm the issue and find a solution09:53
matttin gerrit, can i watch projects based on pattern rather than adding each in individually?09:53
odyssey4mewinggundamth but if you're in a hurry you can 'git tag -l', then 'git checkout -b <tag number>' the latest 12.0.x tag, then execute the playbooks09:54
odyssey4mewinggundamth notes that you will have to at the very least re-execute the repo-build09:54
odyssey4memattt not to my knowledge, which is why I used the review dashboard in the topic :)09:54
*** javeriak has quit IRC09:57
*** javeriak has joined #openstack-ansible09:58
*** electrofelix has joined #openstack-ansible10:02
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone: [WIP] Implement base configuration for independent repository  https://review.openstack.org/28666010:03
openstackgerritMerged openstack/openstack-ansible-os_swift: Implement base configuration for independent repository  https://review.openstack.org/28669210:05
*** javeriak has quit IRC10:08
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_neutron: Implement E501 flake8 check with longer line limit  https://review.openstack.org/28707210:08
winggundamthodyssey4me: I just got what you talk about the branch after I check and seeing your branching system now :)10:08
openstackgerritMerged openstack/openstack-ansible-os_swift_sync: Implement base configuration for independent repository  https://review.openstack.org/28669610:09
*** lynn_kong is now known as Larrie10:12
*** Larrie is now known as larrie10:12
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone: [WIP] Implement base configuration for independent repository  https://review.openstack.org/28666010:12
*** javeriak has joined #openstack-ansible10:15
openstackgerritMatt Thompson proposed openstack/openstack-ansible-os_neutron: Set max-line-length and remove E501 ignore  https://review.openstack.org/28707910:16
odyssey4mehahaha mattt I already did that https://review.openstack.org/28707210:22
matttodyssey4me: ARGH :)10:23
matttabandoned mine :)10:23
odyssey4meheh, great minds think alike :p10:24
*** larrie is now known as larrie_kong10:28
winggundamthdo you know how to replace all openstack repository with user_variables.yml? I put var like requirements_git_repo but it end up clone requirements 2 times. one is main repo and another is the one I put10:31
winggundamthMy company is lack of international bandwidth so I have to mirror repository on our gitlab and want openstack ansible clone from there instead of the main one10:32
*** thorst_ has joined #openstack-ansible10:38
odyssey4mewinggundamth in your user_variably.yml in /etc/openstack_deploy you need to set the *_git_repo vars from https://github.com/openstack/openstack-ansible/blob/master/playbooks/defaults/repo_packages/*.yml to the host you want to git clone from, and ensure that your gitlab server has all those repositories10:39
odyssey4memattt http://logs.openstack.org/60/286660/5/check/gate-openstack-ansible-os_keystone-dsvm-ansible-functional-trusty/1759d25/console.html#_2016-03-02_10_31_08_434 is the failure in https://review.openstack.org/286660 - I need to go afk for a while, if you have a gap you can continue to update the patch to make it work10:40
winggundamthodyssey4me: yes I already did that but it end up cloning 2 times.10:40
odyssey4mewinggundamth hmm, that sounds like a bug10:41
winggundamthoddyssey4me: please see output here http://paste.openstack.org/show/488923/10:42
winggundamthyou can see it clone git.openstack.org and git.example.com twice on the same repo10:42
winggundamthI think maybe because of this so it cause error that mattt trying to reproduce now also10:43
winggundamthbecause it end up cloning and check out on master branch10:44
matttodyssey4me: sure10:45
*** thorst_ has quit IRC10:45
*** hlkv6_ is now known as hlkv610:48
winggundamthI checkout 12.0.6 and doing setup_infrastructure.yml still error at the same task but difference error now http://paste.openstack.org/show/488929/11:01
*** phiche has quit IRC11:03
*** phiche has joined #openstack-ansible11:03
matttwinggundamth: my build has almost finished11:04
matttwinggundamth: so let me understand something11:04
matttwinggundamth: you did a build off liberty, then presumably did a git pull on your openstack-ansible checkout some time later?11:04
matttwinggundamth: if so, you will need to re-run the repo-build.yml play to rebuild all your wheels, venvs, etc.11:05
matttwinggundamth: once that has been done, you can re-run the openstack plays11:05
matttwinggundamth: ah, i see it was your repo-build run that failed :(11:06
winggundamthyeah11:06
winggundamthmattt: please see my latest error from 12.0.611:07
matttwinggundamth: what commit is your branch on ?11:07
winggundamthmattt: 972b41a061be353ca7f44ddc4e3702aae3433a5d11:08
winggundamthI'm sorry but I'm changing to 12.0.6 now11:08
matttwinggundamth: ok go back to liberty branch11:09
matttadn tell me what commit ?11:09
winggundamth3e07fb28a649f2c3e32c8338200d4a21f63c082511:10
matttwinggundamth: yeah that's the head, which i'm on also11:10
matttconfused why my repo-build ran fine11:10
winggundamth:(11:11
winggundamthdo you know the command for that task in ansible?11:11
matttwinggundamth: well what did you run?11:11
winggundamthso I can manually run it in container and hope I can see any useful error11:11
matttwinggundamth: setup-infrastrucutre?11:11
winggundamthyes11:11
matttwinggundamth: the specific play is repo-build.yml, try that11:11
winggundamthsure11:12
winggundamthrunning it now11:14
winggundamthstill the same :(11:22
winggundamthCollecting oslo.i18n<1.6.0,>=1.5.0 (from keystone)11:22
winggundamth  Could not find a version that satisfies the requirement oslo.i18n<1.6.0,>=1.5.0 (from keystone) (from versions: 3.1.0)11:22
winggundamthNo matching distribution found for oslo.i18n<1.6.0,>=1.5.0 (from keystone)11:22
winggundamthIgnoring indexes: https://pypi.python.org/simple11:22
winggundamthI suspect about it said Ignoring indexes11:23
odyssey4meis that during the venv build?11:23
odyssey4meyeah, it looks like it11:24
odyssey4meit's supposed to ignore pypi by that stage11:24
odyssey4meit should already have all the packages compiled into wheels by then11:24
winggundamthic11:24
matttwinggundamth: this is a liberty commit from today, which passed fine: http://logs.openstack.org/91/271591/5/check/gate-openstack-ansible-dsvm-commit/c7ee2a2/console.html11:25
odyssey4meif you cd into /opt/openstack-ansible/playbooks, then run 'ansible repo_all -m shell -a "ls -al /var/www/repo/links/oslo.*' what does it result in?11:25
odyssey4mewinggundamth also, you mentioned that you have an internal git repo - have you implemented any other changes to cater for local isolation?11:26
matttit sounds like something isn't jiving11:26
winggundamthodyssey4me: http://paste.openstack.org/show/488935/11:27
matttodyssey4me: just finished up looking at something for asettle, did you still want me to poke that review ?11:27
* asettle appears11:27
matttwinggundamth: your containers not syncing is also an issue, those files should be the same across the lot11:27
matttwinggundamth: from my deploy which just finished: http://paste.openstack.org/show/488936/11:28
winggundamthmattt: should I rerun playbook from the setup-hosts.yml?11:29
matttwinggundamth: no, let's figure this out11:29
matttwhy you do not have oslo.i18n11:30
winggundamthokay11:30
winggundamththis is my user_variables.yml http://paste.openstack.org/show/488937/11:31
winggundamththe others is default11:31
*** tiagogomes has quit IRC11:34
winggundamthmattt: I can see from ls command that I have /var/www/repo/links/oslo.i18n-3.1.0-py2.py3-none-any.whl11:35
matttwinggundamth: ah right, weird wrapping on that paste11:35
odyssey4meah, so you do have it - which likely means that the issue is that the repo sync hasn't happened11:36
winggundamthmattt: but from the error why it said oslo.i18n<1.6.0,>=1.5.0 not only oslo.i18n>=1.5.011:36
odyssey4mein the container that has the packages, restart the lsync service11:36
odyssey4mewinggundamth also, are your git sources up to date - and are you sure that you have them all?11:37
winggundamthodyssey4me: what do you mean git sources up to date?11:37
odyssey4mewinggundamth you're mirroring the git.o.o repositories for global-requirements, nova, heat, keystone, etc right? (in gitlab) - please make sure that they are all there and up to date11:38
*** furlongm_ has quit IRC11:38
matttwinggundamth: which paste has the oslo.i18n error in it ?11:38
winggundamthodyssey4me: oh yes. I'm sure I have it all. I'm just update it last few hours ago11:38
odyssey4mewinggundamth you should have all the repositories in https://github.com/openstack/openstack-ansible/blob/liberty/playbooks/defaults/repo_packages/openstack_other.yml and https://github.com/openstack/openstack-ansible/blob/liberty/playbooks/defaults/repo_packages/openstack_services.yml11:39
*** furlongm_ has joined #openstack-ansible11:39
winggundamthodyssey4me: yes. I'm sure11:39
winggundamthmattt: http://paste.openstack.org/show/488940/11:40
winggundamthmattt: here you go11:40
matttmaybe you didnt' update your ceilometer sources?  :)11:40
winggundamthwhat I really curious is it asking for <1.6.0 that it shouldn't11:41
odyssey4mewinggundamth it gets requirements from the service git source11:41
winggundamthbecause I have 3.1.0 in my repo container right?11:41
winggundamthic11:41
odyssey4mewinggundamth but all requirements are taken into account in the build process up-front, so that seems like you have something out of date11:41
odyssey4memattt I think that https://review.openstack.org/284896 has to merge before master can be unblocked - it's busy being rechecked as it keeps hitting timeouts11:42
winggundamthand it is the same version that mattt successful build it too11:42
matttwinggundamth: https://github.com/openstack/ceilometer/blob/510aa7897a46688b9826ac8800455de03fe3dc34/requirements.txt#L1811:42
mattti have a feeling your ceilometer source may be out of date11:42
matttcan you verify what you have matches that?11:43
winggundamthsure11:43
odyssey4memattt perhaps check the corresponding upper-constraints too?11:43
*** thorst_ has joined #openstack-ansible11:43
winggundamthmattt: http://paste.openstack.org/show/488942/ this is my requirements.txt from celilometer11:44
odyssey4mewinggundamth there's a set of *.txt files in your repo server under /var/www/repo/os-releases/<tag>/11:45
*** tiagogomes has joined #openstack-ansible11:45
odyssey4mepastebin the contents of those please11:45
winggundamthodyssey4me: this is from 12.0.7 tag https://www.dropbox.com/sh/knhehz5r9nt3usa/AADDJvoUgMw0HiG2ynkIbUhMa?dl=011:48
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: IRR - Implemented for os_keystone  https://review.openstack.org/27971011:48
*** tiagogomes has quit IRC11:50
odyssey4meso global requirements has oslo.i18n>=1.5.0 and constraints has oslo.i18n<=3.1.011:50
*** tiagogomes has joined #openstack-ansible11:50
odyssey4methat's why you have 3.1.0 in your repo11:50
*** thorst_ has quit IRC11:51
winggundamthodyssey4me: yes same as mattt11:51
*** Xu_Chao has joined #openstack-ansible11:51
winggundamthodyssey4me: but what hit me on the wall is why the error show constraints <1.6.011:51
winggundamthodyssey4me: I couldn't find that constraint anywhere11:51
odyssey4meyeah, that's weird11:52
winggundamthodyssey4me: maybe I have to reverse this line with_dict: local_packages.results.0.item.role_packages to see where it get that packages from?11:54
*** furlongm_ has quit IRC11:54
odyssey4mewing perhaps try executing 'openstack-ansible -vv repo-build.yml' so that you can see the commands being executed and their output11:54
*** furlongm_ has joined #openstack-ansible11:55
winggundamthodyssey4me: already done that with -vvvv but nothing useful at all11:56
odyssey4mewinggundamth check the versions of pip, wheel and setuptools across the repo_all group and the hosts group11:56
odyssey4mewheel will come out of 'pip freeze'11:56
odyssey4mepip has a version option I think11:56
odyssey4mesetuptools you may have to find the folder on the fs11:56
winggundamthodyssey4me: http://paste.openstack.org/show/488944/11:57
winggundamthodyssey4me: this is output from -vvvv11:57
odyssey4mewinggundamth I see that https://review.openstack.org/284247 got added after that tag - try checking back out to the head of the liberty branch11:59
odyssey4methen execute the repo-install.yml playbook11:59
winggundamthodyssey4me: this is version http://paste.openstack.org/show/488945/12:01
winggundamthodyssey4me: let me checkout that commit12:02
odyssey4memattt remember this? https://github.com/openstack/openstack-ansible/commit/81f45653fa6bba4bc63fc44785b422b44f321a1712:03
odyssey4mewe haven't yet sorted it all out in liberty12:03
*** sdake has joined #openstack-ansible12:03
matttsadly i do, but that's not the issue here is it?12:03
odyssey4meI think that the wheel is being built in a way that pip can't properly interpret12:04
odyssey4meit's a guess12:04
odyssey4meand it is weird that we're not seeing this in our gate builds12:04
winggundamthmattt odyssey4me do you think if I remove all the containers and rerun whole script again can fix this?12:05
winggundamthalso I actually use deploy host as the same as one management host. is that can cause the problem?12:12
*** admin0 has quit IRC12:12
logan-Hey winggundamth it is not using your local repos from your gitlab due to the trailing .git on the url. Add something like "cinder_git_package_name: cinder" to all of your override definitions12:12
*** sdake has quit IRC12:13
winggundamthlogan: I'm not sure I'm understand. so you suggest me to append '.git' to every gitlab repository? and add *_git_package_name: variable to all repository?12:15
*** admin0 has joined #openstack-ansible12:16
logan-No you already have .git at the end of the repo urls right? (I think gitlab requires this where github does not). This trailing .git in the url confuses repo build and if you look in /var/www/repo/openstack git you will see there are 2 copies of all the repos12:16
winggundamthlogan-: exactly right12:17
winggundamthlogan-: so what do you suggest me to do?12:17
logan-Specify the package name manually and it will override correctly instead of trying to clone both12:17
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Use current, but pinned versions of pip, setuptools and wheel  https://review.openstack.org/28716112:17
logan-So right below the line where you override with your gitlab url, put that package name var as well12:18
winggundamthok12:18
odyssey4melogan- ah, well spotted :) you've clearly been through this12:18
winggundamththat's clearly12:19
*** neilus has joined #openstack-ansible12:20
odyssey4memattt I've done https://review.openstack.org/287161 pre-emptively - it's a backport of the two patches, one of which has been voted through but hasn't merged yet thanks to shenanigans. I expect that we should get this patch in before the next tag to save a lot of people pain and suffering.12:21
odyssey4mebut I expect it could wait for the next tag, assuming that people keep up to date :p12:21
matttlogan-: so is that the source of winggundamth's problems?12:22
*** jguy_ has quit IRC12:22
matttodyssey4me: cool, i'd love to see the master patch merge first12:24
*** asettle has quit IRC12:24
odyssey4memattt me too, of course -that's why I WIP the patch :)12:24
matttwho knows at this rate if it will get through :P12:25
odyssey4memattt https://review.openstack.org/284896 is on a rax node, so it's got a good chance of getting through - that should unblock master12:25
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: IRR - Implemented for os_keystone  https://review.openstack.org/27971012:29
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: IRR - Implemented for all os_* and dep roles  https://review.openstack.org/28498712:29
*** javeriak has quit IRC12:30
*** subscope has joined #openstack-ansible12:30
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Use current, but pinned versions of pip, setuptools and wheel  https://review.openstack.org/28716112:34
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Use current, but pinned versions of pip, setuptools and wheel  https://review.openstack.org/28497712:35
*** subscope has quit IRC12:36
odyssey4memattt that's set dependencies, so hopefully the builds will succeed and we can get them in sooner12:37
odyssey4mealright, now I'm going to be afk for an hour or two - bbl!12:37
logan-mattt: maybe... I am not sure on the broken requirement but I would think if the same shas are being used the requirements should be calculated the same regardless of there being 2 copies of all the repos.12:38
logan-i just saw that exact error last week when I accidentally had a neutron stable/kilo sha in my liberty install :)12:38
*** asettle has joined #openstack-ansible12:39
logan-so i would suggest also doing grep -R '<1.6.0,>=1.5.0' /var/www/repo/openstackgit12:39
logan-on the repo server12:39
winggundamthlogan-: already did that and find nothing :(12:41
logan-winggundamth: example overrides file btw http://cdn.pasteraw.com/vt6p3tbaaii28jpmn8pfd8rfk1pevz12:42
winggundamthrerun repo-build again. bless me12:42
winggundamthlogan-: thank you. will try that12:42
logan-you might want to rm /var/www/repo and rebuild from scratch to clean out the mess there if this repo-build run still fails12:43
*** subscope has joined #openstack-ansible12:43
*** subscope has quit IRC12:44
winggundamthlogan-: yeah already done that ;)12:45
*** subscope has joined #openstack-ansible12:46
*** thorst_ has joined #openstack-ansible12:48
*** markvoelker has quit IRC12:48
*** markvoelker has joined #openstack-ansible12:49
*** fawadkhaliq has joined #openstack-ansible12:54
winggundamthbravo!!!12:58
winggundamththank you so much logan- odyssey4me mattt. now I passed the task I have problem12:59
*** woodard has joined #openstack-ansible13:00
*** woodard has quit IRC13:00
*** woodard has joined #openstack-ansible13:01
logan-great!13:01
winggundamthanother error T_T13:05
*** jguy_ has joined #openstack-ansible13:05
winggundamthbut i'm fine. will try to find the way to fix it first13:05
*** johnmilton has joined #openstack-ansible13:07
winggundamthhttp://paste.openstack.org/show/488952/ :(13:08
logan-ls /var/www/repo/pools/mysql_python13:08
*** johnmilton has quit IRC13:08
winggundamthhttp://paste.openstack.org/show/488953/13:08
*** johnmilton has joined #openstack-ansible13:09
winggundamthroot@tdc-ops-mgt-01:/openstack/infra1_repo_container-5de053b6/repo/pools/mysql_python# ls13:09
winggundamthmysql_python-1.2.5-cp27-none-linux_x86_64.whl13:09
logan-make sure lsyncd is running on the primary repo container where it ran the build13:09
matttwinggundamth: so what was the problem?13:10
winggundamthmattt: it repo name that gitlab need to append with .git so it will create folder append with .git also13:10
matttwinggundamth: :(  is there something we can do in osa to help with this situation since it burned you and logan- ?13:11
winggundamthmattt: I'm thinking about another session for who want to speed up deploy by caching like git, pypi, apt-cacher-ng and rackspace repo13:13
logan-last year when cloudnull was working on the repo build stuff i hit that issue immediately.. so he added the _package_name override. it fixes it of course but I am not sure if it is documented anywhere.. having a little doc about how to override repos would be helpful i think13:13
winggundamthand I'm sure that we need more document about variable here13:13
winggundamthmaybe if I can pass all of these I can help on document for you because I need heavily using cache from the bw problem13:14
winggundamthlogan-: http://paste.openstack.org/show/488956/13:17
winggundamthlogan-: they don't have 12.0.7 in os-release http://rpc-repo.rackspace.com/os-releases/13:17
matttany reason why we couldn't just strip .git off the repo name ?13:17
logan-that's strange. pip_lock_down should set pip to use your local repo server13:17
matttfeels like this should be handled internally without overriding vars13:17
winggundamthmattt: that's another clever way to do it!13:18
logan-mattt: yeah I don't disagree. as a deployer I would expect it to handle a git url with trailing .git in stride :)13:19
matttcool, i'll have a look at this13:19
winggundamthlogan-: maybe because of this var in my user_variables.yml http://paste.openstack.org/show/488961/13:20
winggundamthany variable that will overrid this?13:20
logan-i would avoid overriding openstack_repo_url13:21
*** fawadkhaliq has quit IRC13:21
logan-openstack_repo_url: "http://{{ internal_lb_vip_address }}:{{ repo_server_port }}"13:22
logan-that is pointing at your repo containers by default13:22
winggundamthgrrr ic13:22
winggundamththat's not clear in document also13:22
winggundamthhttp://docs.openstack.org/developer/openstack-ansible/install-guide/configure-hostlist.html13:23
* winggundamth rerun again13:24
*** furlongm_ is now known as furlongm13:26
*** Bjoern_ has joined #openstack-ansible13:28
*** skamithi has joined #openstack-ansible13:31
*** rewriteurl has joined #openstack-ansible13:33
*** Bjoern_ has quit IRC13:34
*** cemmason1 has quit IRC13:38
*** cemmason has joined #openstack-ansible13:38
logan-mattt odyssey4me: it does look like rpc-repo is missing 12.0.6 wheels btw13:41
*** mathlin1 has joined #openstack-ansible13:52
*** retreved has joined #openstack-ansible13:56
matttcloudnull: it looks like https://github.com/openstack/openstack-ansible-os_keystone/commit/34ad63b41eb85bc947bb51e3d3895b7516374943 got unwound in your os_keystone update, was that deliberate ?14:00
*** cemmason has quit IRC14:03
odyssey4melogan- rpc-repo is manually built, and RPC is not yet using liberty - so yeah, it may or may not have built wheels on it, but it's only there to obtain wheels if they happen to be there - pypi is still accessed as well and wheels/sdists can come from there14:04
*** cemmason has joined #openstack-ansible14:04
*** bapalm has quit IRC14:05
odyssey4memattt FYI https://review.openstack.org/287205 will be needed for https://review.openstack.org/284987 to pass14:05
*** bapalm has joined #openstack-ansible14:06
matttodyssey4me: voted14:06
*** weshay has joined #openstack-ansible14:07
*** cemmason1 has joined #openstack-ansible14:08
winggundamtherror still keep coming ;( http://paste.openstack.org/show/488976/14:09
*** busterswt has joined #openstack-ansible14:09
*** cemmason has quit IRC14:09
winggundamthI remember that when I doing this with aio. I can pass it instantly :(14:09
matttwinggundamth: are you running memcache ?14:10
odyssey4methanks mattt14:10
winggundamthinfra1_memcached_container-0b5296e7            RUNNING  10.0.3.73, 172.27.5.224                 -     YES (onboot, openstack)14:10
winggundamthinfra1_memcached_container-380ae605            RUNNING  10.0.3.245, 172.27.5.151                -     YES (onboot, openstack)14:10
winggundamthmattt: yes I think14:10
*** KLevenstein has joined #openstack-ansible14:10
odyssey4memattt after a good run on a rax nodepool instance, https://review.openstack.org/284896 is doing the gate check on ovh and is going far too slowly :/14:16
*** Xu_Chao has quit IRC14:16
winggundamththis is detail from -vvvv http://paste.openstack.org/show/488979/14:17
*** Xu_Chao has joined #openstack-ansible14:19
*** subscope has quit IRC14:20
*** subscope has joined #openstack-ansible14:21
*** subscope has quit IRC14:21
*** subscope has joined #openstack-ansible14:22
matttwinggundamth: which playbook are you running btw, and are you adding any --limits ?14:25
mhaydenis there some gate brokenness around virtualenv?14:25
matttmhayden: master presumably ?14:26
winggundamthmattt: no --limits. at first I'm running this openstack-ansible setup-infrastructure.yml and after that I added --tag rabbitmq-ssl14:26
mhaydenmattt: yessir14:26
matttwinggundamth: try removing the tag, why are you using that out of interest?14:26
winggundamthmattt: I'm running without --tag first and got the error so I don't want to run the whole step again so I use --tag14:27
*** subscope has quit IRC14:27
winggundamthmattt: I'll try ansible-playbook rabbitmq-install.yml14:28
matttwinggundamth: you can just run the rabbitmq portion of setup-infrastrucure.yml14:28
automagicallyShould this be marked WIP until functional tests are complete? https://review.openstack.org/#/c/286650/14:28
matttwinggundamth: yep14:28
odyssey4memhayden master is currently quite broken - you'll need to apply https://review.openstack.org/284896 then https://review.openstack.org/279710 then https://review.openstack.org/284987 and also https://review.openstack.org/284977 to make it all go again properly14:28
matttautomagically: naw14:28
mhaydenodyssey4me: haha, wow14:29
odyssey4memhayden you'll notice that those patches are already chained up14:29
mhaydenimpressive14:29
matttmhayden: you're welcome :)14:29
winggundamthmattt: no luck. still the same error :(14:30
mhaydeni'm looking at the lbaasv2 commit -- i assume i'll need to wedge that into the os_neutron IRR ?14:30
odyssey4meautomagically nope, first get linters passing and the gate enabled - then evaluate whether any patches went in after the code was extracted into a seperate repo - then work on implementing a working functional check14:30
odyssey4meautomagically see https://etherpad.openstack.org/p/openstack-ansible-role-breakout :)14:30
automagicallyThx, will start in on the repo diff14:30
winggundamthmattt: I curious about error "Memcache key not found"14:30
openstackgerritMerged openstack/openstack-ansible-os_swift_sync: Add empty role dependency list  https://review.openstack.org/28720514:30
odyssey4memattt I've checked https://review.openstack.org/286650 and it seems fine - but was hoping that you'd check it before I vote :)14:31
matttwinggundamth: i'm grasping at straws here, but you could try shutting down 2 memcache containers to make sure it's not something with that14:31
matttodyssey4me: looking14:31
winggundamthmattt: and this is error from -vvvv 'The key you specified was not found within memcached. If you are load balancing memcached, attempt to connect to a single node.'14:31
winggundamthmattt: ok will try14:32
*** jthorne has joined #openstack-ansible14:34
matttautomagically: one small issue w/ that review, otherwise looks GTG14:34
automagicallymattt: Thx for catching that, will amend14:34
matttnp14:34
odyssey4meautomagically jmccrory cloudnull a quick review of https://review.openstack.org/287072 would be great14:35
matttwe'll have to go through and standardise some of the descriptions and READMEs in the roles, they're all over the place :)14:35
matttbut a task for another day14:35
odyssey4me++14:35
*** shausy has quit IRC14:36
odyssey4meI'd actually like each role's docs to have details of how the service is setup, what it expects and does and how to configure it all... then the install guide can focus on simple stuff and refer to the role docs for more details14:36
automagicallyodyssey4me: That sounds like a good goal, but definitely down the road14:37
mhaydenhow do we handle release notes for the IRR roles?14:37
automagically^ That there is a great question. I believe there are some already in the main repo for changes in the IRRs14:38
winggundamthis it usual that it has 2 containers for the same role per host?14:38
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-os_horizon: Implement base configuration for independent repository  https://review.openstack.org/28665014:39
mhaydenautomagically: okay14:40
*** scarlisle has joined #openstack-ansible14:41
*** mgoddard_ has joined #openstack-ansible14:44
openstackgerritMerged openstack/openstack-ansible-os_glance: Implement base configuration for independent repository  https://review.openstack.org/28666714:46
Bofu2Uhmm. How hard would it be to factor in the swift3 middleware into the playbooks?14:46
openstackgerritMerged openstack/openstack-ansible-os_aodh: Implement base configuration for independent repository  https://review.openstack.org/28665514:46
openstackgerritMerged openstack/openstack-ansible-os_heat: Implement base configuration for independent repository  https://review.openstack.org/28669314:46
openstackgerritMajor Hayden proposed openstack/openstack-ansible-os_neutron: Replace LBaaS v1 with v2  https://review.openstack.org/28723714:47
odyssey4memhayden the only instrumentation at this time for release notes is in the main repo - we build all docs there14:47
*** b3rnard0 has quit IRC14:47
mhaydenodyssey4me: got it14:47
openstackgerritMerged openstack/openstack-ansible-os_neutron: Implement E501 flake8 check with longer line limit  https://review.openstack.org/28707214:47
odyssey4memhayden as we build out the documentation in the independent repositories, I expect that we'll also need to implement release notes there14:47
*** mgoddard has quit IRC14:47
*** subscope has joined #openstack-ansible14:48
odyssey4memhayden but in this cycle I think it's best that we do it in the main repo14:48
matttwinggundamth: sorry not familiar w/ that error, i'd have to dig through the plugin and playbooks to see what is causing that14:48
*** admin0 has quit IRC14:48
* odyssey4me shakes a stick at ovh nodepool instances14:48
*** b3rnard0 has joined #openstack-ansible14:48
cloudnullmorning14:49
winggundamthmattt: no worry. I have to go now too. almost 10pm here14:49
Bofu2Umornin cloudnull14:49
matttwinggundamth: sorry, please come back tomorrow and we can try this again :)14:49
cloudnullohai Bofu2U14:49
winggundamthmattt: I'm sure that we will talk again tmr :)14:49
*** alejandrito has joined #openstack-ansible14:49
matttwinggundamth: great, have a good evening14:50
cloudnullBofu2U: how goes the kernel experiment ?14:50
Bofu2Uthe the 4x series still up14:50
Bofu2Uthe 3.18 lasted about 3-4x as long, then died14:50
winggundamththank you so much for all your help. I'm really appreciate it.14:50
matttwinggundamth: thank logan- , he solved your problem14:50
cloudnullBofu2U: well thats good news. kinda :)14:50
mattti'd never have guessed that in a million years :P14:51
*** sdake has joined #openstack-ansible14:51
*** admin0 has joined #openstack-ansible14:51
*** Mudpuppy has joined #openstack-ansible14:51
Bofu2Ucloudnull yeah good enough I guess14:52
*** Mudpuppy has quit IRC14:52
*** b3rnard0 has quit IRC14:52
*** b3rnard0 has joined #openstack-ansible14:52
*** Mudpuppy has joined #openstack-ansible14:52
*** b3rnard0 is now known as Guest1112614:53
*** Guest11126 has quit IRC14:53
*** b3rnard0- has joined #openstack-ansible14:54
openstackgerritMajor Hayden proposed openstack/openstack-ansible: Replace LBaaS v1 with v2  https://review.openstack.org/28724314:55
cloudnullBofu2U:  so  upgrade all the things to 4.x and all will be right with the world :)14:56
Bofu2Uallegedly ;)14:56
cloudnullBofu2U: RE: swift3 we dont have a role for it14:56
cloudnullso we'd have to make that14:56
Bofu2Uah ok14:56
Bofu2Uyeah im just trying to get deis setup with the swift store in general, ill try to do it direct first14:56
cloudnullseems like https://github.com/openstack/swift3can be added to the existing swift role as it'd have to be added as middleware14:57
cloudnull*https://github.com/openstack/swift3 can14:57
odyssey4meif it's just additional middleware I expect it wouldn't be too hard14:57
odyssey4mewhat is swift3?14:58
* odyssey4me is out of touch14:58
*** dmsimard is now known as dmsimard|afk14:58
Bofu2Ubasically makes swift work with the S3 API14:58
cloudnullits s3 compat within swift -- it new bits that have to be installed on the proxy nodes and middleware added. so it shouldnt be too hard to do14:59
*** mgoddard_ has quit IRC14:59
cloudnullbut i dont have the cycles to look at it right now.14:59
*** mgoddard has joined #openstack-ansible14:59
cloudnullBofu2U:  if you have some time to work on that i'd be happy to review it14:59
Bofu2Urgr14:59
Bofu2Ugonna try to get this with swift direct, but if it doesn't work I'll try to slate some time for that15:00
*** skamithi13 has joined #openstack-ansible15:00
cloudnullsweet!15:00
* cloudnull bbl driving in15:00
odyssey4meyeah, I do think that we should add the replacement S3 and EC2 API add-ons into the stack as an optional extra15:01
odyssey4methe newer stuff is a lot better, I hear, and it'd be a nice feature for those who want it15:01
*** spotz_zzz is now known as spotz15:03
automagicallyodyssey4me: Any objections to changing the tox.ini in os_keystone IRR to run with -vvv or similar. Gate is failing and I’m seeing no useful output in http://logs.openstack.org/60/286660/5/check/gate-openstack-ansible-os_keystone-dsvm-ansible-functional-trusty/1759d25/console.html other than “failed: [keystone1] => {"attempts": 5, "failed": true, "parsed": false}”15:04
odyssey4meautomagically I already added the human readable output - you're not going to get much better than that15:04
automagicallyugh15:04
odyssey4meautomagically the only way to figure that one out is going to be to look at a running build15:05
automagicallyLooks like its our keystone library blowing up15:05
odyssey4meautomagically or that the task can't reach the service, either because the service isn't running or because the task is being given the wrong info15:05
automagicallyk, thx, will dig in a bit further, just not used to seeing such limited output in the logs15:06
odyssey4meautomagically a suggestion - do what we've done for the other roles to make it effectively pass the functional test... let's get the base stuff merged and make the functional test work in a subsequent patch15:06
automagicallyWill do15:06
*** sigmavirus24_awa is now known as sigmavirus2415:08
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-os_keystone: Implement base configuration for independent repository  https://review.openstack.org/28666015:09
*** ShannonM has joined #openstack-ansible15:11
*** javeriak has joined #openstack-ansible15:11
*** javeriak has quit IRC15:12
*** javeriak_ has joined #openstack-ansible15:16
*** raddaoui has joined #openstack-ansible15:21
*** winggundamth has quit IRC15:21
*** izaakk has joined #openstack-ansible15:23
*** skamithi13 has quit IRC15:25
*** skamithi13 has joined #openstack-ansible15:26
*** weezS has joined #openstack-ansible15:28
*** weezS has quit IRC15:28
openstackgerritMatt Thompson proposed openstack/openstack-ansible-plugins: Strip .git from _git_repo variables  https://review.openstack.org/28726615:31
*** rewriteurl has quit IRC15:34
*** rsparule has quit IRC15:40
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-lxc_container_create: Reformat YAML condition to list  https://review.openstack.org/28727315:45
openstackgerritMatt Thompson proposed openstack/openstack-ansible-plugins: Strip .git from _git_repo variables  https://review.openstack.org/28726615:47
*** keedya has joined #openstack-ansible15:49
*** sdake_ has joined #openstack-ansible15:51
*** sdake has quit IRC15:52
*** mathlin1 has quit IRC15:52
NepocI looked at this just at the right time... I need the strip .git from _git_repo variables today15:52
openstackgerritMajor Hayden proposed openstack/openstack-ansible-os_neutron: Replace LBaaS v1 with v2  https://review.openstack.org/28723715:53
*** Mudpuppy has quit IRC15:54
*** Mudpuppy has joined #openstack-ansible15:54
*** sdake has joined #openstack-ansible15:55
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update new user secrets in Liberty  https://review.openstack.org/28684415:56
openstackgerritMerged openstack/openstack-ansible-pip_install: Test version of pip being installed  https://review.openstack.org/28528415:56
*** eric_lopez has quit IRC15:57
*** sdake_ has quit IRC15:58
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update new user secrets in Liberty  https://review.openstack.org/28684416:00
*** admin0 has quit IRC16:00
*** pcaruana has quit IRC16:05
*** rromans has quit IRC16:06
*** fawadkhaliq has joined #openstack-ansible16:06
*** rromans has joined #openstack-ansible16:08
*** sdake_ has joined #openstack-ansible16:12
*** flaviosr has joined #openstack-ansible16:13
*** fawadkhaliq has quit IRC16:14
*** alextricity25_ has joined #openstack-ansible16:14
*** odyssey4me_ has joined #openstack-ansible16:15
*** bryan_att has quit IRC16:15
*** larrie_kong has quit IRC16:15
*** alextricity25 has quit IRC16:15
*** larrie_kong has joined #openstack-ansible16:15
*** thorst_ has quit IRC16:15
*** maximov_ has quit IRC16:15
*** odyssey4me has quit IRC16:15
*** sdake has quit IRC16:15
*** git-harry has quit IRC16:15
*** kencjohnston has joined #openstack-ansible16:15
*** larrie_kong has quit IRC16:15
*** larrie_kong has joined #openstack-ansible16:15
*** git-harry has joined #openstack-ansible16:15
*** thorst_ has joined #openstack-ansible16:16
*** maximov_ has joined #openstack-ansible16:17
*** phalmos has joined #openstack-ansible16:18
*** bryan_att has joined #openstack-ansible16:19
*** sdake_ is now known as sdake16:21
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update ceilometer environment configuration  https://review.openstack.org/28682716:21
*** rromans has quit IRC16:21
*** rromans has joined #openstack-ansible16:23
*** git-harry has left #openstack-ansible16:24
palendaeraddaoui, izaakk: Upgrade stuff is at https://review.openstack.org/#/q/topic:liberty_upgrade feedback on whether the docs are clear or not would be great16:26
tiagogomeswhy is there " user: root" in the playbooks?16:26
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-galera_server: Cleanup configuration for independent repository  https://review.openstack.org/28730016:26
*** daneyon has joined #openstack-ansible16:28
*** phalmos has quit IRC16:28
*** daneyon has quit IRC16:28
*** daneyon_ has joined #openstack-ansible16:29
izaakkthanks palendae , will help there16:29
*** phalmos has joined #openstack-ansible16:30
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Add aodh, haproxy, and neutron config upgrades  https://review.openstack.org/28675316:32
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update ceilometer environment configuration  https://review.openstack.org/28682716:32
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update new user secrets in Liberty  https://review.openstack.org/28684416:32
*** daneyon_ has quit IRC16:33
openstackgerritMajor Hayden proposed openstack/openstack-ansible: Replace LBaaS v1 with v2  https://review.openstack.org/28724316:33
*** daneyon has joined #openstack-ansible16:33
*** odyssey4me_ is now known as odyssey4me16:33
*** electrofelix has quit IRC16:34
palendaeFor anyone interested in liberty upgrades: https://review.openstack.org/#/q/topic:liberty_upgrade. https://review.openstack.org/#/c/286251/ is the base commit that everything else relies on, fwiw16:34
*** keedya has quit IRC16:34
*** sdake has quit IRC16:35
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Add missing runtime information to upgrade script  https://review.openstack.org/28625116:35
palendaeArgh, sorry odyssey4me, just realized that commit needed a rebase after you +W :(16:36
odyssey4mepalendae no worries, a rebase shouldn't lose the votes16:36
palendaeCleared the +W16:36
palendaeAnd the Jenkins vote16:37
* palendae grumbles16:37
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Add upgrade config change tools  https://review.openstack.org/28556416:37
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Add aodh, haproxy, and neutron config upgrades  https://review.openstack.org/28675316:38
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update ceilometer environment configuration  https://review.openstack.org/28682716:38
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update new user secrets in Liberty  https://review.openstack.org/28684416:38
*** eric_lopez has joined #openstack-ansible16:38
* palendae backs off the rebase button for a bit16:38
*** eric_lopez has quit IRC16:40
*** rcarrillocruz has left #openstack-ansible16:40
*** sdake has joined #openstack-ansible16:40
*** eric_lopez has joined #openstack-ansible16:40
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-openstack_openrc: No longer defining OS_TENANT_NAME in openrc  https://review.openstack.org/28730916:42
*** cemmason1 has quit IRC16:43
*** cemmason has joined #openstack-ansible16:43
*** jwitko has joined #openstack-ansible16:44
*** subscope has quit IRC16:44
*** daneyon_ has joined #openstack-ansible16:47
odyssey4mepalendae I'm told by infra that the check and gate queues always rebase before testing - so we actually don't have to rebase if we depend on the head of the branch.16:49
palendaeAh16:49
*** daneyon has quit IRC16:49
palendaePart of that is so some of the nits that exist in parent or grandparent change for upgrades get funneled down16:49
palendaeCause some of these depend on each other now, rather than directly on HEAD16:50
spotzpalensae poke if I need to re-review anything if the votes are lost16:50
palendaespotz: No +1s were lost, just Jenkins and +W; the rebasing thing is I'm just spamming the channel :p16:51
spotzhehe16:51
palendaeBut https://review.openstack.org/#/q/topic:liberty_upgrade if I'm screwing up my wordings16:51
odyssey4mecloudnull I'm kinda thinking of temporarily dropping affinity in master to get those IRR patches through, then we can up the affinity again. We're sitting in the crappy situation where all the gate checks go through OVH which starts the smoke test at 89mins and therefore the job times out17:01
odyssey4meunless you have any other ideas which we can implement in the repositories to make them go faster?17:01
odyssey4meautomagically logan- jmccrory mattt ^ ?17:01
*** subscope has joined #openstack-ansible17:02
cloudnullodyssey4me:  seems sensible17:03
cloudnullespecially while ovh is being a pain17:03
odyssey4mewell, the trouble is that ovh will never speed up - so I'd far rather we put some optimisations in than rely on reducing testing17:04
odyssey4meultimately that's the more sustainable route17:04
odyssey4mecloudnull we need to get https://review.openstack.org/286660 in as a starting point for os_keystone, then try to implement the functional testing in a working state afterwards17:06
logan-yeah that makes sense.. what are the pain points on these ovh instances? is it network latency from mirrors? io bound? wondering what optimizations would be helpful17:08
odyssey4melogan- not entirely sure - we do know that we have the option of using the mirrors which infra have implemented... but they aren't signed, so we need extra instrumentation for that and it has to follow an integrated master patch anyway, so that's no good17:09
odyssey4mewe have to figure out how to do some optimisations in the roles themselves, because those gate every time and gate quicker17:10
odyssey4methey then have a downstream effect on the integrated testing17:10
odyssey4methere may be no option except to reduce affinity in order to open the door though, let me see what I can come up with - but any ideas would be great17:10
*** electrofelix has joined #openstack-ansible17:10
odyssey4methere are logs in http://logs.openstack.org/96/284896/14/gate/gate-openstack-ansible-dsvm-commit/9f4b6d5/ which capture the state of the instance in various places during the build which may help17:11
odyssey4mespecifically http://logs.openstack.org/96/284896/14/gate/gate-openstack-ansible-dsvm-commit/9f4b6d5/logs/instance-info/17:12
odyssey4mestevelle any thoughts on options?17:13
*** sdake has quit IRC17:13
stevelleI'm fine with reduced affinity for the moment, again, remembering we did this about 6 months ago.17:14
odyssey4meyeah we did17:15
logan-i am thinking about pulling down a bunch of review logs and trying to parse out the play recap stuff and attribute averages with different providers to see % diffs between them17:15
logan-it would help to know which tasks are taking so much longer on ovh and see whether they are io intensive/cpu intensive, etc and look at ways to optimize from there17:16
odyssey4melogan- I've just checked some logs and saw that in some ovh regions we don't even get to the point where the os_nova role has completed before it times out.17:16
*** sdake has joined #openstack-ansible17:17
logan-just looking thru 1 of the random ovh logs i noticed nova venv took nearly 2 minutes to extract... seems high?17:17
odyssey4mebut yeah, specifically check the gate-openstack-ansible-dsvm-commit logs in https://review.openstack.org/#/c/284896/17:17
logan-k17:17
*** phiche has quit IRC17:18
odyssey4meyou'll have to check the whole history of logs to find them - you'll see that on other providers we usually get through just fine17:18
odyssey4mebut in the commit job specifically is where we most often land on ovh and other troublesome providers17:18
odyssey4mebut when I last looked through the job results using logstash.openstack.org I found that 50% of failures were coming from two ovh regions - one in the us and one in europe17:19
*** fawadkhaliq has joined #openstack-ansible17:21
*** subscope has quit IRC17:22
*** Bjoern_ has joined #openstack-ansible17:23
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Reduce galera/rabbit/horizon/keystone/repo affinity to 1  https://review.openstack.org/28733117:23
palendaespotz: Mind looking at https://review.openstack.org/#/c/286753/11/doc/source/upgrade-guide/scripts.rst, L28? Trying to decide if the verb should be 'add' or 'adds'. I think 'add' is correct now17:23
odyssey4melet's see if https://review.openstack.org/287331 reduces the check by enough time to matter17:24
*** electrofelix has quit IRC17:25
*** Gayathri has joined #openstack-ansible17:26
*** spotz is now known as spotz_zzz17:26
*** skamithi has quit IRC17:27
GayathriHi17:27
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-openstack_openrc: Adding Vagrantfile for local developer testing  https://review.openstack.org/28733317:27
Gayathriwhether the openstack-ansible community playbook support OVS?17:27
Gayathrifor liberty17:27
*** raddaoui has quit IRC17:27
*** raddaoui has joined #openstack-ansible17:28
*** raddaoui_ has joined #openstack-ansible17:28
openstackgerritAndy McCrae proposed openstack/openstack-ansible-os_swift: Fix Swift log directory ownership  https://review.openstack.org/28733717:29
openstackgerritAndy McCrae proposed openstack/openstack-ansible-os_swift: Fix Swift log directory ownership  https://review.openstack.org/28733717:31
*** galstrom_zzz is now known as galstrom17:32
*** raddaoui_ has quit IRC17:33
*** raddaoui has quit IRC17:33
*** raddaoui has joined #openstack-ansible17:33
odyssey4meGayathri not in liberty, nor in master (yet) - there is work ongoing to make that happen with the hope that it lands for Mitaka, but it may only make Newton17:33
odyssey4meautomagically nice idea in https://review.openstack.org/287333 :) we should do that everywhere17:35
automagicallyodyssey4me: Cool, hope its easy enough to do. I’d rather not continue to crush OVH infra if it never helps us and I’d likewise like to run higher affinities and catch more bugs on all the other Zuul infra17:36
odyssey4meautomagically also, can we add some basic docs to help developers prep their laptop in order to actually use vagrant for the testing?17:37
palendaeKind of related to that, I'm using this for my test AIOs now https://github.com/nrb/osad-basic-test-node. I may add a vagrantfile17:37
automagicallyodyssey4me: Whoops, thought you were referencing my comment on: https://review.openstack.org/#/c/287331/117:37
palendaeThough I'm doing that repo for myself, not the general public17:37
*** spotz_zzz is now known as spotz17:37
automagicallyBut as for the vagrant stuff, I’ll add some docs in the Contributor section17:37
*** dmsimard|afk is now known as dmsimard17:38
odyssey4meautomagically yeah, it'd be great to actually have a few vagrantfiles to use on server hardware for gate tests - it'll help anyone who wants to implement a multinode test quickly17:38
odyssey4mebut that'll be for the future :)17:38
automagicallyBelieve we’ve got an outstanding review with a Vagrant based AIO17:38
automagicallyThat looks pretty exciting17:38
odyssey4meautomagically we can actually detect the provider and fail fast if we want to, but if we do that we will get nothing through the integrated gate as I think what tends to happen is that a provider is used almost exclusively for some types of tests from time to time17:39
odyssey4meright now we have a slight chance that it may get through - and we do get some data to try and isolate the issues17:40
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-openstack_openrc: Correct filename of clouds.yml to clouds.yaml  https://review.openstack.org/28734217:40
automagicallyodyssey4me: Thanks for the additional gate infra context17:40
odyssey4meautomagically yeah, I think it might be a good idea for me to share how we use OpenStack-CI at the summit... and how we're supposed to use it. :) I have discussed actually having some of the infra folk in a session like that so that they can provide some guidance with regards to how we can do better to optimise what happens in the gate and to reduce failures.17:42
odyssey4meAt the very least pabelanger might join us, if not some others. :)17:42
automagicallyodyssey4me: I’d definitely be interested in learning more. I’m also trying to explore whether Comcast can provide some Nova compute infra for Zuul17:42
palendaeodyssey4me: That'd be good; I get the sense that we're actually pushing infra to its limits17:43
palendaeI'm not sure infra was meant to actually deploy an integrated openstack17:43
odyssey4meautomagically ah nice!17:43
*** sdake has quit IRC17:43
*** raddaoui has quit IRC17:43
odyssey4meone of the topics that I think we also want to chat to pabelanger about is his roles for doing CI in your own infrastructure using OpenStack-CI technology... he's done a great job of pulling that infra together and I know that CI/CD is a topic many of us are interested in exploring17:44
Gayathrithanks for the clarification odyssey4me17:44
logan-are there roles for that somewhere? i had no idea17:45
odyssey4melogan- take a look through https://github.com/openstack?query=ansible-role and https://github.com/openstack/windmill17:46
logan-cool thanks17:46
odyssey4meI'd like to ask for him to give us an overview of how it fits together, what it does and how it can be used... it'd be great if we could adopt this as a subproject, assuming we're all interested in participating in the future development17:47
palendaeodyssey4me: For upgrade testing that'd be cool, since I highly doubt we'll get that within a 90 minute limit in ~12 months17:47
palendaeDeploying 1 version, then upgrading to a 2nd means roughly double the time17:48
pabelangerodyssey4me: Yup, have no issues explaining how it works or even demoing it17:48
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-openstack_openrc: [DOCS] Cleanup the role docs for consistency and clarity  https://review.openstack.org/28734717:49
*** raddaoui has joined #openstack-ansible17:49
odyssey4mepalendae yeah, I expect that the upgrade tests will need to be done as a periodic job rather than an on-commit job... although there is also the possibility of external CI which, if using appropriate hardware, could give fast enough feedback to make it worthwhile17:50
palendaeodyssey4me: Yeah. I've also toyed with the idea of adding smaller tests around the scripts that do config changes17:50
palendaeIt's admittedly a small part of the overall work, but still more than we have now17:51
odyssey4mepalendae that's a good idea17:52
odyssey4mepalendae also note that within the roles upgrades can be tested specifically for the role, potentially, and the turnaround time there is much shorter17:52
pabelangerpalendae: that's something I'd like to test with some of the ansible-roles I have. First deploy HEAD~1, ensure things work, then deploy HEAD and recheck.  From a role POV, don't expect it to take long.17:52
pabelangermoving it into periodic would be a good solution too17:52
palendaeYeah, for roles, definitely. But the integration will still need to be tested17:53
*** electrofelix has joined #openstack-ansible17:53
pabelangeragreed17:53
*** electrofelix has quit IRC17:53
jmccroryodyssey4me logan- i put a script together the other day to parse through logs and spit out timings of per task and per playbook, i'll put that up somewhere today17:53
*** asettle has quit IRC17:53
jmccrorycontainer creates took twice as long on ovh compared to rax17:54
stevellejmccrory: nice17:54
logan-nice. i was just pulling a list of log files to download and do that lol17:54
jmccroryso affinity should help somewhat17:54
logan-are the flavor specs significantly different between the providers?17:54
stevelleI started one a couple weeks ago but never finished it17:54
*** Gayathri has quit IRC17:55
*** raddaoui has quit IRC17:56
*** kencjohnston has quit IRC17:58
odyssey4melogan- all the flavors have the same cpu/ram spec - but different disk specs18:01
odyssey4methe host contention and the underlying hardware (cpu, ram, disk) will be different for every provider18:02
logan-sure, if venv extracts and container creates are taking a long time, it makes you wonder about single threaded cpu performance and io bottlenecks18:02
* odyssey4me is downloading his first vagrant box to test out automagically's patch :)18:02
*** git-harry has joined #openstack-ansible18:04
*** git-harry has left #openstack-ansible18:04
*** git-harry has joined #openstack-ansible18:05
*** phalmos has quit IRC18:07
*** skamithi has joined #openstack-ansible18:08
jwitkohey odyssey4me, how are you today18:08
openstackgerritMerged openstack/openstack-ansible-os_horizon: Implement base configuration for independent repository  https://review.openstack.org/28665018:08
*** skamithi13 has quit IRC18:08
*** skamithi13 has joined #openstack-ansible18:09
*** skamithi has quit IRC18:09
*** mgoddard has quit IRC18:11
jwitkoHaving an issue I'm wondering if you could help me fix... i think its a quick one.    My route table on my containers is not being set with a proper gateway for the management interface.  http://paste.openstack.org/raw/489018/  -- as you can see in that route table there is no default gateway set for 10.96.76.0/24 (it should be 10.96.76.1).  My interface is defined in the user config18:12
jwitkoas follows, with the gateway specified  http://paste.openstack.org/raw/489019/  .  Any idea what I'm doing wrong?18:12
*** keedya has joined #openstack-ansible18:12
*** sigmavirus24 is now known as sigmavirus24_awa18:12
* odyssey4me looks18:12
*** willlo has joined #openstack-ansible18:15
*** mikelk has quit IRC18:17
*** cemmason has quit IRC18:21
*** cemmason has joined #openstack-ansible18:21
*** skamithi13 has quit IRC18:21
jwitkoodyssey4me, do I need to add the gateway at the same level as the management_netmask variable?18:23
jwitkoit looks like the template checks if gateway is defined at that level18:23
odyssey4mejwitko as per http://docs.openstack.org/developer/openstack-ansible/install-guide/configure-networking.html#adding-static-routes-to-network-interfaces do you see the container having the /etc/network/interfaces.d/ with the post-up command?18:24
jwitkoodyssey4me, yes I do.  it has the correct gateway in the via18:24
jwitkobut its not coming up with that gateway18:24
jwitkoif I add a static route to the gateway manually, it works just fine18:25
odyssey4mejwitko have you tried adding it manually inside the container to see if it works?18:25
jwitkoodyssey4me, yes and adding it manually makes the container work perfectly18:26
jwitkoit can respond to ICMP without any issues once I add it18:26
odyssey4mejwitko and then if you remove it and ifdown/ifup it doesn't work?18:27
odyssey4meand are there any entries in syslog that may pertain to the actions?18:27
jwitkowell I added it via the route command, not the interface file18:27
*** subscope has joined #openstack-ansible18:28
odyssey4mejwitko ok, but the lxc-container-create adds it into the interface file - so is it in the interface file?18:28
jwitkoodyssey4me, here is the interface file18:28
jwitkohttp://paste.openstack.org/raw/489024/18:29
jwitkoit adds the "via gateway" on the 2nd to last line18:29
jwitkobut that is not actually setting it as a default gateway for eth118:29
jwitkoso when it recieves packets from the outside it is looking to respond via its only default gateway, the 10.0 internal one.  which is of course on a different interface so it drops it18:30
odyssey4mejwitko the eth0 interface has a default gateway too - if you remove that, then ifdown eth1/ifup eth1 - does it work?18:33
*** jwagner is now known as jwagner_lunch18:35
*** zhangjn has quit IRC18:37
jwitkothe other one is dhcp18:37
odyssey4mejwitko sure - I'm asking for that to be done as a test18:38
jwitkoeth0, is dhcp18:38
jwitkoyou mean remove it using the route command ?18:38
odyssey4meyep18:38
*** zhangjn has joined #openstack-ansible18:38
jwitkosure one moment18:38
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [WIP] Gate: Set pip index and extra-index to OpenStack-CI repositories  https://review.openstack.org/26586718:39
jwitkoodyssey4me, I have removed it but I'm getting a very odd error on eth118:40
jwitkoifdown: interface eth1 not configured18:40
jwitkothe interface is up, it has an IP18:40
odyssey4mejwitko have you restarted the container since adding the extra route?18:41
jwitkoI just restarted it before18:41
odyssey4mejwitko was the container created first, then the route added later - or was it all done at the same time?18:41
*** sdake has joined #openstack-ansible18:41
jwitkoso, in its initial created form the route table looks as I pasted above18:42
jwitkohttp://paste.openstack.org/raw/489018/18:42
odyssey4mejwitko so honestly I have no idea what could be wrong - perhaps Apsu or rackertom could assist?18:42
odyssey4mesomething appears to be a bit weird there18:42
*** alejandrito has quit IRC18:43
odyssey4meit seems that you've done the config right from an OSA standpoint, but the interface isn't doing what's expected of it18:43
jwitkoodyssey4me, if I want to delete all the containers and rebuild them18:43
odyssey4meI'm really crappy at networking18:43
jwitkoI just need to delete my inventory.json generated file and run the lxc-container-detroy right ?18:43
jwitkoor vice versa18:43
odyssey4mejwitko you could try and destroy one of the containers and rebuilt it to see if it works better18:43
odyssey4mejwitko you can leave the entry in the inventory18:43
odyssey4meno need to regenerate the inventory at all18:43
jwitkowell I'm thinking I want to regenerate the inventory with the "gateway: " variable inside the container interface config18:44
odyssey4mejust execute the lxc-container-detroy playbook using the container name as a limit18:44
odyssey4meah ok, well you can destroy the container, then remove it from the inventory using inventory-manage (in the scripts folder), then yes you can recreate the whole thing18:44
ApsuWhat's up?18:45
ApsuOh, ifdown is complaining.18:45
jwitkoHey Apsu,  here is the original issue:18:45
jwitkoHaving an issue I'm wondering if you could help me fix... i think its a quick one.    My route table on my containers is not being set with a proper gateway for the management interface.  http://paste.openstack.org/raw/489018/  -- as you can see in that route table there is no default gateway set for 10.96.76.0/24 (it should be 10.96.76.1).  My interface is defined in the user config18:45
jwitkoas follows, with the gateway specified  http://paste.openstack.org/raw/489019/  .  Any idea what I'm doing wrong?18:45
ApsuYeah, ifup/ifdown are Ubuntu's way of managing interface configuration. When ifdown says that, it just means ifup wasn't used to configure it.18:46
ApsuReading18:46
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [WIP] Gate: Set pip index and extra-index to OpenStack-CI repositories  https://review.openstack.org/26586718:46
odyssey4meI wonder if ip set dev down / ip set dev up implements the post-up commands?18:48
*** subscope has quit IRC18:48
Apsujwitko: First of all, don't use route or ifconfig. They're abandoned and fundamentally broken.18:48
*** jthorne has quit IRC18:48
*** subscope has joined #openstack-ansible18:48
Apsujwitko: Second... you can't have two (working) default gateways. The default gateway isn't supposed to be your management network18:48
ApsuIt's supposed to be the LXC container network, on eth0. Which you have, according to the first paste.18:49
*** subscope has quit IRC18:49
jwitkoApsu, ok but then why are packets getting dropped going out ?18:49
jwitkoto me it looks like they're trying to go out the default gateway, on eth0 after they are received on eth1.18:49
spotzautomagically you still changing the clouds.yaml one?18:49
ApsuElaborate. Which packets, generated from what, going to where, and how are you determining they're dropped?18:49
jwitkoApsu,  so packets are coming from an F5 load balancer via icmp health check going to the container IP address.18:50
ApsuContainer management IP?18:50
jwitkoI am determining they're dropped by listening for all/any ICMP on the local IP and I only see the incoming ICMP requests, nothing going out18:50
jwitkoApsu, correct.18:50
jwitkosorry, container SSH IP18:51
jwitkoI don't know if thats what manage is, or if the 10.0.3 is the mgmt18:51
ApsuThat's not clarifying my question. According to your paste configs, 10.96.76.0/24 is your management CIDR18:51
ApsuOn eth1 in the container.18:51
jwitkoyes, correct18:51
ApsuSo you're getting traffic to that IP from your F5, yes?18:51
jwitkothen yes, container management IP.18:51
*** retreved has quit IRC18:52
ApsuOk. What IP is that traffic coming from?18:52
jwitkoit comes from a self-IP on the load balancer.   10.96.112.518:52
ApsuSo there's your problem. It's not in your management CIDR.18:52
ApsuSo when the container receives it, it doesn't have a route back to that CIDR, so it uses the default gateway18:52
jwitkoexactly18:53
ApsuIf your F5 is going to talk to your containers, you need them to be on the same CIDR, ideally.18:53
jwitkobut the default gateway for the vlan of the mgmt network is 10.96.97.118:53
jwitkobut thats not really true,  if the containers could just have their default gateway for eth1 set then they can talk just fine18:53
ApsuThat's only because you've got routing that makes it work on the host side, where your .1 lives.18:54
ApsuOr did you put .1 on the F5?18:54
jwitkono, .1 is on a cisco device18:54
ApsuThen that's where the routing that can see the F5's self-ip is.18:54
ApsuEither way, it's only due to that it will work.18:54
ApsuNot because what you're suggesting is more correct.18:54
*** admin0 has joined #openstack-ansible18:55
ApsuThe simpler and more correct way would be to use an IP in the same range to do your health checks from :P18:55
*** admin0 has quit IRC18:55
jwitkoyea..18:55
*** admin0 has joined #openstack-ansible18:55
ApsuSee, the containers need to hit the internet for various things18:55
jwitkothat requires adding the vlan to the F5 interface, adding a self-ip to the F518:55
ApsuAnd we setup the host's lxc bridge to allow that, by default18:56
ApsuWhich is plumbed into containers on eth018:56
ApsuSo that's why they get the default gateway18:56
*** subscope has joined #openstack-ansible18:56
ApsuChanging that default gateway would require that your management network can hit the internet, instead18:56
ApsuSo if that's not how you already have it setup... it's going to break other things18:56
jwitkowell,  the containers use a proxy for everything18:56
ApsuThat's fine.18:57
jwitkoalright.  it makes sense now18:58
jwitkothank you for your time and explanation18:58
ApsuCool. You bet.18:58
jwitkoApsu, one more question if you don't mind18:58
ApsuSure18:58
jwitkoIn researching this I just noticed the ability to set MTUs specifically for SAN interfaces in the user config file18:58
jwitkomy inventory.json is already generated, is there any way for me to keep all the same container IPs and stuff (they're already added to my load balancer) but get that bit added into the inventory file ?18:59
*** galstrom is now known as galstrom_zzz18:59
*** Bjoern_ is now known as Bjoern_zZzZzZzZ18:59
ApsuYes. Changing things in the user config and rerunning the appropriate playbooks won't generate new IPs for the containers, it will merge in new config18:59
ApsuYou can test this yourself if you're nervous by backing up your inventory and just running the inventory generator itself manually, and inspecting the output19:00
*** thorst_ has quit IRC19:00
ApsuIt's a python file19:00
jwitkoah ok cool,  what do I run to regenerate the json ?19:00
*** thorst has joined #openstack-ansible19:01
*** thorst has quit IRC19:01
*** thorst has joined #openstack-ansible19:01
*** skamithi13 has joined #openstack-ansible19:01
ApsuUnder playbooks/inventory is https://github.com/openstack/openstack-ansible/blob/master/playbooks/inventory/dynamic_inventory.py19:01
odyssey4meexecuting /opt/openstack-ansible/playbooks/inventory/dynamic_inventory.py should just spit out the json19:01
odyssey4meto stdout19:02
jwitkothanks guys19:02
ApsuLooks like we've got a --list arg too.19:02
ApsuGuessing that may pretty-print :P19:02
odyssey4meApsu you may wish to inspect https://review.openstack.org/#/c/286078/ and suggest alternative methods that could be used :)19:02
*** Bjoern_zZzZzZzZ is now known as Bjoern_19:03
*** phiche has joined #openstack-ansible19:05
*** jthorne has joined #openstack-ansible19:06
*** jthorne has joined #openstack-ansible19:06
jmccrorylogan- stevelle odyssey4me https://gist.github.com/jimmymccrory/312542d5f56d14b35f9d19:06
*** weezS has joined #openstack-ansible19:06
Apsuodyssey4me: Ugh.19:07
*** jwagner_lunch is now known as jwagner19:09
ApsuSo many better ways to do it. Like using the --check feature. And || to catch the error.19:10
odyssey4meApsu review away :)19:10
odyssey4mejmccrory AttributeError: 'NoneType' object has no attribute 'group' when run against console.html :/19:11
jmccroryoh yeah...doesn't work directly against the html, needs to be the text copied out of it19:11
odyssey4menow if only I could point it at a URL ;)19:12
*** kencjohnston has joined #openstack-ansible19:12
jmccroryhah yeah, can work on that. more regex19:13
logan-thanks jmccrory that looks a lot less crappy than the one i just made19:13
automagicallyspotz: Yep, no more changes to the cloud.yaml review19:13
*** flaviosr has quit IRC19:13
logan-i just pulled down 2 gigs of the console.htmls and im regexing against that. it is pretty slow but getting some initial stuff gathered up now19:13
spotzokie going to look:)19:13
automagicallyGoing AFK for a bit. Gotta eat some lunch19:15
odyssey4melogan- jmccrory FYI: http://logstash.openstack.org/#dashboard/file/logstash.json?query=project%3A%5C%22openstack%2Fopenstack-ansible%5C%22%20AND%20message%3A%5C%22Build%20timed%20out%20(after%2090%20minutes).%20Marking%20the%20build%20as%20failed.%5C%2219:18
*** abitha has joined #openstack-ansible19:20
stevelleovh, timeouts.  Seems like a fair bit of wasted compute because of that timout. Maybe some clouds should have a longer timeout.19:21
odyssey4mestevelle it's unfortunately a one size fits all19:21
odyssey4meand infra won't let us increase the timeout because we're behaving badly19:22
*** sigmavirus24_awa is now known as sigmavirus2419:22
odyssey4mewe don't use their apt repositories, or their pip config, or any of their cached git repositories19:22
palendaestevelle: Yeah, odyssey4me has been saying for a while that we're not really playing by infra's rules, so they're not willing to increase that yet19:23
odyssey4meso basically we need to optimise, use the facilities they provide - or we need to switch to using external CI for integrated gating19:23
Apsuodyssey4me: Reviewed.19:23
odyssey4methanks Apsu19:24
ApsuYep19:25
palendaeSeems like using the cached git repos would be low hanging fruit19:25
palendaeNot sure that it would be the biggest time saver since I've not benchmarked, though19:26
*** weezS has quit IRC19:26
odyssey4mewell, we also do other things like use external repositories for galera, a downloaded deb for rabbit, etc19:27
odyssey4meall these are moving parts which go outside of the cloud provider and therefore cause higher latency and potential failures19:28
palendaeRight19:28
stevellepalendae: the cached git repos bit should be low hanging and if it isn't slower than it gets us closer to infra's good side which is enough19:28
palendaeSome of those are inherent risks19:28
odyssey4metheir apt repositories right now are only ubuntu, but I believe they're open to adding other stuff19:29
stevellefaster doesn't need to be the outcome in that regard19:29
stevellehelps if it is19:29
odyssey4meso the trouble is that the cached stuff is all on the host, so we'd have to figure out how to (only in the gate) bind mount that into the repo container and consume the cached stuff19:29
palendaeAhhh19:29
*** eil397 has joined #openstack-ansible19:30
palendaeYeah, so you'd need to bind mount it and provide a local git url19:30
odyssey4meall their optimisations are built with devstack in mind19:30
palendaeRight19:30
palendaeHence 'DevStack VM' (dsvm)19:30
odyssey4methe issue I have with all this is that we deviate in the gate quite substantially from a real world deployment19:30
palendaeRight19:31
palendaeI was gonna say :)19:31
palendaeI mean, an AIO is already a substantial deviation19:31
palendaeAnyone really using that other than for sanity checks is asking for trouble19:31
odyssey4mebut we may have to do something like that for on-commit gating and then have a periodic job (or external CI) which doesn't deviate as much19:31
odyssey4meyes, this is true19:31
cloudnullodyssey4me:  https://review.openstack.org/#/c/287331/  seems to work19:32
*** galstrom_zzz is now known as galstrom19:32
cloudnullovh env in 1 hour 20 min19:32
odyssey4meI'm not too much of a fan of creating code paths specifically for OpenStack-CI usage, but it seems we may have to19:32
odyssey4mecloudnull aha, that's a good result19:32
odyssey4menow the trouble is that it relies on the repo server adjustments which won't merge without that19:33
logan-http://cdn.pasteraw.com/hclu0giwlnb01b4ir36er4p5nr3mbwn19:33
*** albertcard has joined #openstack-ansible19:33
odyssey4memaybe we'll have to combine the patches?19:33
odyssey4me*repo build adjustments19:33
*** weezS has joined #openstack-ansible19:33
odyssey4meso I'm suggesting that we combine https://review.openstack.org/#/c/284896/ and https://review.openstack.org/#/c/287331/ to unblock the master gate19:34
palendaeodyssey4me: Nor am I. I wonder what the puppet project does for their integration19:35
odyssey4methoughts? stevelle cloudnull automagically jmccrory19:35
odyssey4mepalendae they only added an integration test recently, and I see them make it non voting often19:35
stevelleodyssey4me: abandon 287331, merge with parent, modify commit msg of parent to reference 287331.19:35
odyssey4mestevelle yep, that's my thinking19:35
palendaeodyssey4me: Gotcha.19:36
jmccrorysounds good to me19:36
logan-data points used for that: http://cdn.pasteraw.com/qmj1wpl527x93jnh3at8xp2whrwszcn19:36
palendaelogan-: Awesome\19:37
odyssey4melogan- any way we can get a reverse sorted result in reverse order of longest to shortest?19:37
jmccrorynice logan-19:37
odyssey4meeven just a top 20 will do19:37
odyssey4meexcellent extraction by the way, thanks jmccrory and logan-19:37
stevellelogan-: based on that I think using the apt caches and git repo caches seems highly likely to make ovh faster, ths19:37
odyssey4mewow: lxc_container_create | Create container:19:38
odyssey4medevstack-trusty-rax-iad (43 samples) - 199.5619:38
odyssey4medevstack-trusty-ovh-bhs1 (58 samples) - 429.1319:38
*** retreved has joined #openstack-ansible19:38
palendaeDang19:38
logan-yeah one sec @ sorted19:38
palendaeOver twice as long19:38
logan-local mirror usage would help a lot but there are still things that are super slow on ovh that don't use any network like extracting things etc19:38
odyssey4merepo_build | Clone all upstream git repositories:19:39
odyssey4medevstack-trusty-rax-iad (43 samples) - 254.2519:39
odyssey4medevstack-trusty-ovh-bhs1 (58 samples) - 331.8519:39
*** raddaoui has joined #openstack-ansible19:39
*** skamithi has joined #openstack-ansible19:39
palendaeI can't remember, does the create container task hit anything hosted by rackspace?19:39
stevelleI still see 5 minutes in low-hanging fruit19:39
stevelleeven if lxc doesn't get faster19:39
odyssey4mepalendae only this one:19:40
odyssey4melxc_hosts | Download lxc cache(s):19:40
odyssey4medevstack-trusty-rax-iad (43 samples) - 10.9119:40
odyssey4medevstack-trusty-ovh-bhs1 (58 samples) - 20.2919:40
jmccroryi tested async/polling locally against the container create task over the weekend but didn't see much improvement unfortunately, not sure what can be done there19:40
*** retreved has quit IRC19:40
palendaeSo still twice as long, but an order of magnitude less that the create19:40
*** retreved has joined #openstack-ansible19:40
logan-yea all of the unarchive venv tasks are about the same % diff19:41
odyssey4mea little improvement here could do well: lxc_hosts | Prepare cached image:19:41
odyssey4medevstack-trusty-rax-iad (43 samples) - 44.9719:41
odyssey4medevstack-trusty-ovh-bhs1 (58 samples) - 107.1219:41
logan-it seems like ovh is just very slow to extract things19:41
palendaeIs OVH still 8GB RAM?19:41
odyssey4meyeah, I suspect that rax is using ssd's while ovh is using spinning disks19:41
stevellepalendae: it is19:41
odyssey4meyep all flavors have the same cpu8/ram819:41
palendaeOk19:41
logan-if there are more cores available maybe it would help to use a threaded untar19:41
palendaeThought so but worth confirming19:41
*** jwagner is now known as jwagner[away]19:42
stevellereasonable to guess ovh is i/o bound19:42
odyssey4meall the same - I've already done the investigation19:42
logan-:(19:42
odyssey4meand infra won't provide different flavors, even if they have access to them19:42
palendaeSure19:42
odyssey4methey don't want to set a precedent19:42
palendaeThat's reasonable19:42
odyssey4mealso, I asked about using two nodes instead of 1 - they're open to it, once we've become a good citizen and have proved that 1 node isn't enough19:43
odyssey4methey also warned that the failure rate for two nodes is typically higher19:43
*** phalmos has joined #openstack-ansible19:43
palendaeSo some of this needs to be done even if it's not necessarily going to speed things up19:43
odyssey4mein our setup, I don't really see two nodes improving performance anyway19:43
stevelleonly if we decrease the density of containers really19:44
palendaeYeah, unlikely for performance19:44
stevellebut then the added overhead19:44
palendaeDifference between us and the rest of openstack is we're trying to emulate a full cluster19:44
odyssey4methe repo build still only happens in one container19:44
palendaeNot just spinning up the bare minimum19:44
odyssey4methe lxc cache prep would happen in the same way on both hosts19:44
logan-http://cdn.pasteraw.com/7hv2xz5dopjwlf0qi0iu87lx7ohqfoe sorted19:46
*** phalmos has quit IRC19:47
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Implement repo_build IRR & Reduce galera/rabbit/horizon/keystone/repo affinity  https://review.openstack.org/28489619:49
stevelleany chance that zfs would have any benefit on ovh for the storage bottleneck19:49
*** galstrom is now known as galstrom_zzz19:50
odyssey4mestevelle cloudnull jmccrory automagically https://review.openstack.org/284896 should unblock master to give us some breathing room19:50
odyssey4mestevelle not sure - what're you thinking?19:50
odyssey4meI know that I've thought that perhaps we should change the way we do the file-based container image prep19:51
odyssey4memaybe use overlayfs or something19:51
stevellewondering if we can do overlayfs or something on ovh-only to get speed. I know we didn't see an improvement on ssds19:51
odyssey4methat way we wouldn't have to extract the same data so many times19:51
stevelleI was thinking a copy on write might help19:51
odyssey4meit would be an overhaul of the way we prep the container image though19:52
palendaeI think hughsaunders looked at that kind of thing in Juno and saw a performance decrease19:52
odyssey4mebut hey - it could cut the time taken to create the containers fairly substantially19:52
palendaeBut maybe it was during container runs19:52
palendaeDon't remember the details, I just know he tried some COW FS19:52
stevellepalendae: that was also on an ssd backend19:52
stevellewe weren't i/o bottlenecked19:52
openstackgerritMajor Hayden proposed openstack/openstack-ansible-os_neutron: Add LBaaS v2 support  https://review.openstack.org/28723719:52
odyssey4mepalendae yeah, but when I asked all he did was have the fs be overlayfs... in our container prep we still do a tarball extract of the data into the fs of the container19:53
odyssey4mewell, we did back then19:53
odyssey4methe process has changed somewhat and it may be worth the experiment again19:53
palendaeYeah19:53
palendaeAh, right19:53
cloudnullodyssey4me stevelle we wont be able to gain anything with the overlayfs bits if the kernel isnt 3.19+19:53
odyssey4meI was working on a revised method of prepping the containers in https://review.openstack.org/272743 - I haven't had the time to get it done19:54
palendaeHe just changed the fs type in the lxc configs19:54
stevellewondering if the tarball is actually faster on ovh, maybe not19:54
stevellecloudnull: that's right, recalling that now19:54
palendaecloudnull: Also good point, but we're also moving towards 16.04 support "soon"19:54
stevellenot soon enough for that sadly19:54
odyssey4mestevelle ++19:54
cloudnull++19:54
cloudnullif we convert to the download template its quite a bit faster.19:54
palendaeIs 14.04 still < 3.19?19:54
cloudnullpalendae:  3.13 by default19:55
palendaeBah19:55
cloudnull4.2 is available19:55
odyssey4mepalendae yeah, you need to add the wily kernel package to upgrade it19:55
palendaeYeah, so upgrade and reboot19:55
palendaeBleh19:55
odyssey4mecorrect19:55
stevelleback to liberty upgrades (boo)19:55
openstackgerritBjoern Teipel proposed openstack/openstack-ansible-os_neutron: Fix auth_url for metadata_agent.ini  https://review.openstack.org/28741119:55
* palendae should do that too19:55
*** neilus has quit IRC19:56
stevellemore yak shaving to get toward 3.1919:56
openstackgerritBjoern Teipel proposed openstack/openstack-ansible-os_neutron: Fix auth_url for neutron configuration  https://review.openstack.org/28741119:58
*** sdake_ has joined #openstack-ansible19:59
*** sdake has quit IRC20:00
*** thorst has quit IRC20:01
*** thorst has joined #openstack-ansible20:01
*** hybridpollo has joined #openstack-ansible20:01
*** sdake has joined #openstack-ansible20:03
*** thorst has quit IRC20:05
*** neilus has joined #openstack-ansible20:05
*** sdake_ has quit IRC20:06
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Use current, but pinned versions of pip, setuptools and wheel  https://review.openstack.org/28497720:06
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [WIP] Gate: Set pip index and extra-index to OpenStack-CI repositories  https://review.openstack.org/26586720:06
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: IRR - Implemented for os_keystone  https://review.openstack.org/27971020:06
*** willlo has quit IRC20:07
*** thorst has joined #openstack-ansible20:08
*** galstrom_zzz is now known as galstrom20:09
*** neilus has quit IRC20:10
*** javeriak_ has quit IRC20:10
*** keedya has quit IRC20:19
*** unlaudable has joined #openstack-ansible20:27
*** permalac has quit IRC20:30
*** keedya has joined #openstack-ansible20:31
*** raddaoui has quit IRC20:31
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-openstack_openrc: No longer defining OS_TENANT_NAME in openrc  https://review.openstack.org/28730920:32
odyssey4meautomagically ^ minor change to the commit message - using the change id allows you to click through and see the original commit and any backports20:33
automagicallyAh cool, thanks for that odyssey4me20:33
automagicallycloudnull: You around?20:39
*** mathlin1 has joined #openstack-ansible20:42
cloudnulli am20:44
automagicallycloudnull: Had a qq about the intent of https://github.com/openstack/openstack-ansible-os_keystone/blob/master/tests/stand-alone.yml20:45
automagicallyI _think_ its a test for determining if the role can function outside of the typical OSA inventory patterns, but that isn’t completely clear from the file itself20:46
cloudnullits to test keystone without a db or messaging backend20:46
cloudnullbasically yes to test keystone outside of a typical osa use case20:46
automagicallyOkay, the vars for rabbit and galera definitely obscure the intent20:47
cloudnullLike if you wanted to run keystone with a custom auth provider that didnt require access to a db20:47
automagicallyPerhaps some commentary on the top of the playbook is in order20:47
cloudnullyea. not all of that is finished.20:47
automagicallyAh, gotcha20:47
automagicallyThanks for the context, was digging into the tests there to see if I could be of help in getting the gate functional tests running for that role20:48
odyssey4meI quite like that idea - test.yml can be a meta-playbook that includes other playbooks - but each test playbook should have some sort of description of what it's trying to achieve20:48
odyssey4mealthough, that said, perhaps that info belongs in the docs for the role instead?20:48
cloudnullautomagically:  theres a stand-alone inventory too20:49
stevelleSeems odd to document test cases in docs20:49
cloudnullwhich is called by the tox functional tests .20:49
automagicallySince I’m a TDD guy most of the time, I tend to expect the tests to communicate their intent before I go looking for docs outside of the tests20:49
stevelle+120:49
odyssey4mestevelle perhaps, but I'm kind-of thinking that the docs for the role should describe the use-cases, and then point at the specific test to illustrate how that use case is implemented20:49
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-os_swift: Adding Vagrantfile for local developer testing  https://review.openstack.org/28743020:50
automagicallyodyssey4me: That combination would work well20:50
stevelleI'd be interested to see how that might look20:50
odyssey4meso sure - a short description in the playbook at the top, but the detailed background, purpose, etc in the docs20:50
stevelleit could work20:50
*** dank_ has joined #openstack-ansible20:50
*** jwagner[away] is now known as jwagner20:51
odyssey4meit would be a great way for people to understand how the role was designed to be used, what it expects as input and what it's designed to deliver20:51
automagicallyAnsible role documentation and testing is definitely an evolving art within the community20:51
automagicallyI think an ideal though is very close to what Jesse just described, which is docs that contain example playbook content that then link to the test suite for each example20:52
odyssey4meof course I would absolutely love it if we could use something like docstrings for the vars and have them automatically rendered for the roles in documentation20:52
*** dank_ has left #openstack-ansible20:52
stevelleodyssey4me: I'm looking at it as a way to merge your expressed interest in losing examples and replacing with docs.  I'm seeing tests as the replacement, and now you're shifting how the docs play into that.20:53
odyssey4meautomagically yep, that's pretty much what I'm suggesting - let the docs contain the long explanations of background/purpose and refer to the playbook for implementation details (which could change regularly)20:53
odyssey4mestevelle the trouble with the .example files is that they are documentation - but that documentation is not kept in sync with other documentation20:54
odyssey4meand developers update one or the other20:55
stevelleagreed, if they were exercised as tests, that helps20:55
openstackgerritTravis Truman (automagically) proposed openstack/openstack-ansible-os_swift: Removing unneeded with_items usage for clarity  https://review.openstack.org/28743520:55
stevellebut the melding with a doc of a use case is interesting20:55
*** kencjohnston has quit IRC20:55
stevelleI look forward to seeing PoC20:55
palendaeodyssey4me: fwiw, our upgrade process uses .example files20:55
odyssey4meideally I see the docs as something that should contain stuff that doesn't change much - anything that does change often should ideally be automatically generated20:56
*** b3rnard0- is now known as b3rnard020:56
palendaeActually nevermind, the one I'm thinking of uses a straight .yml file https://review.openstack.org/#/c/286844/8/scripts/upgrade-utilities/playbooks/user-secrets-adjustment.yml20:56
openstackgerritMajor Hayden proposed openstack/openstack-ansible: Replace LBaaS v1 with v2  https://review.openstack.org/28724320:56
odyssey4meso, for instance, the purpose of the role and the use cases it was designed to resolve don't change often - if anything you add more use cases20:57
openstackgerritMajor Hayden proposed openstack/openstack-ansible: Add LBaaS v2 support (OSA)  https://review.openstack.org/28724320:57
odyssey4methe tests are then designed against the use cases, and the code implements the use cases20:57
mhaydenApsu: would you mind giving me a looksee on https://review.openstack.org/#/c/287237/ and https://review.openstack.org/#/c/287243/ when you have time?20:57
odyssey4meso all the docs have to do is refer to the code and tests for implementation details20:58
odyssey4mebut yeah, I need to find some time to put together some sort of PoC of how I see that all working20:58
openstackgerritBjoern Teipel proposed openstack/openstack-ansible: Adding release notes for erlang VM tuning parameters  https://review.openstack.org/28550620:59
*** kencjohnston has joined #openstack-ansible20:59
logan-more profiling from the stuff earlier.. much larger sample set. http://cdn.pasteraw.com/n2bzegt8grah7b5whijxmnyxtukc2v0 http://cdn.pasteraw.com/3l2wxvhdnqsfk6jr3q5snm17wwz5jkb http://cdn.pasteraw.com/5efufwesy1nmxnl6qis9z1zr1hbce020:59
odyssey4mewow, ovh-gra1 took over 5 mins just to clone repositories21:00
Apsumhayden: Sure.21:01
*** skamithi has quit IRC21:01
mhaydenApsu: muchas gracias21:01
automagicallylogan-: That profiling is pretty crazy useful, thanks for putting that together21:02
logan-no prob21:02
odyssey4me++21:02
*** neilus has joined #openstack-ansible21:03
odyssey4meuh, wut:21:03
pabelangerodyssey4me: where did you see that?21:03
odyssey4merepo_build | Remove pool indexes if found:21:03
odyssey4medevstack-trusty-bluebox-sjc1 (50 samples) - 31.9821:03
odyssey4medevstack-trusty-rax-iad (21 samples) - 40.4621:03
odyssey4merepo_build | Make sure old venv build directories are clean:21:03
odyssey4medevstack-trusty-bluebox-sjc1 (25 samples) - 9.8121:03
odyssey4melogan- are these counters the values all added together, or are they averages or aggregates?21:04
pabelangerodyssey4me: I think ovh has reported some network issues recently, but nothing has been confirmed21:04
odyssey4mepabelanger logan-'s pulled together some profiling info for our integrated test run so that we can look for optimisations - here's the set per provider: http://cdn.pasteraw.com/n2bzegt8grah7b5whijxmnyxtukc2v021:05
odyssey4meorganised from longest to shortest task execution21:05
logan-http://cdn.pasteraw.com/j40s1vfmxmuwlm66sbijvn0yafd6m0a averages21:05
Apsumhayden: In the neutron_upstart_init.yml, line 123-124 reference neutron_services['neutron-lbaas-agent'] in the when: conditional, but lines 116-117 reference neutron_services['neutron-lbaasv2-agent']. Am I missing something?21:06
mhaydengimme 5 to look21:07
mhaydenjumped into another bug :/21:07
Apsunp21:07
pabelangerodyssey4me: logan-: interesting, checking21:08
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update user variables to Liberty names  https://review.openstack.org/28743821:08
palendaerromans, erikmwilson When the gate builds the docs, I'd like you guys to look at https://review.openstack.org/#/c/287438/21:09
Apsuafk, grabbing some food21:09
palendaeThat commit has a dynamically build table of variable name changes21:09
palendaebuilt*21:09
pabelangerwould be interested to see which repos repo_build | Clone all upstream git repositorie does21:09
pabelangerlikely could be optimized by using local cache21:10
odyssey4mepabelanger yeah, the trouble is that it's done inside a container21:10
pabelangerodyssey4me: have you done any updates to use the afs mirror?21:10
odyssey4meso we'd have to bind mount the local cache to the container21:10
pabelangerodyssey4me: Ya, container stuff would be interesting for git clones21:10
odyssey4mepabelanger I'm working on it - haven't had time to finish that up - the trouble is going to be configuring all containers to not validate the key21:11
rromanspalendae: kk21:11
pabelangerodyssey4me: cool, would be interested to see that code.  I haven't does anything yet myself21:11
pabelangerI wonder if some sort of lxc container cache could be created in the dib process21:12
odyssey4mepabelanger the big challenge is that for every optimisation we do for the gate, we deviate from real world deployment... and in doing so we don't properly check what a real world deployment would be like21:12
pabelangerso you have some sort of warm cache when jenkins runs21:12
pabelangerodyssey4me: right21:12
odyssey4mealso, implementing special code paths for OpenStack-CI alone just doesn't feel right21:13
palendaeLong shot, but does anyone know if reno releasenotes are aware of the Sphinx docs in the doc directory? I would like to add a link to detailed documentation in my release note, but would like to avoid hard linking to a URL that's not there yet21:13
pabelangersomething like AFS mirrors might be fine, since you are just pinning to the closest one for apt, pypi and soon rpms21:13
odyssey4mebut yeah, we need to do better - so I'm working on figuring out how to go about it without compromising the purpose of the integration test itself21:13
pabelangerlocal git repos, ya, I see what you are saying21:14
odyssey4mepabelanger yeah, we already had implemented 'closest mirror' stuff for apt - we just need to convert that into 'use the host's configured mirror, but allow an unsigned repo'21:14
openstackgerritMajor Hayden proposed openstack/openstack-ansible-os_keystone: Remove dangling Default domain cfg file  https://review.openstack.org/28744021:15
*** unlaudable has quit IRC21:15
odyssey4meusing the host's default mirror is actually how our AIO is designed to work by default, but allowing an unsigned mirror is not something we would ever advocate for production and thus it has never been instrumented21:15
pabelangerodyssey4me: Hmm, I though gpg was working for the mirrors. Would have to test it again21:15
odyssey4meit is? if so, then it's changed and we can undo some of the stuff we implemented to work around that21:16
odyssey4meour security role, for instance, had to skip the test which checks that unsigned repositories were not enabled21:16
pabelangernot sure, I would double check with -infra.  I don't think we are doing anything special to the packages that would require a new gpg key for them21:17
odyssey4mepabelanger they're not doing a straight mirror - I had a bit of a chat with monty about it asking why21:17
odyssey4methey're actually building their own repo21:17
odyssey4meso the indexes for the repo don't match the ubuntu indexes21:18
odyssey4meand therefore you need to allow the repo to be unsigned21:18
openstackgerritMajor Hayden proposed openstack/openstack-ansible: Release notes for dangling domain fix  https://review.openstack.org/28744421:18
mhaydenApsu: good catch on that -- fixing now21:19
odyssey4mepabelanger hence https://github.com/openstack/openstack-ansible/commit/b4bb5105b62404b4f9c5e6f1b0384e5f60b6550621:20
openstackgerritMajor Hayden proposed openstack/openstack-ansible-os_neutron: Add LBaaS v2 support  https://review.openstack.org/28723721:20
mhaydenautomagically: dang, thanks for the fast +1 :P21:20
*** cemmason has quit IRC21:20
*** unlaudable has joined #openstack-ansible21:20
*** cemmason has joined #openstack-ansible21:20
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update user variables to Liberty names  https://review.openstack.org/28743821:20
automagicallymhayden: I remembered the change from OSA, so its a simple review to see you move it to the IRR21:21
mhaydenwoot21:21
mhaydenthanks anyways, automagically  :)21:21
* mhayden needs one more core on openstack-manuals :/21:21
*** unlaudable has quit IRC21:22
*** dnt has joined #openstack-ansible21:23
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update user variables to Liberty names  https://review.openstack.org/28743821:23
pabelangerodyssey4me: Ah, that is right. Forgot about that21:23
*** cemmason1 has joined #openstack-ansible21:23
*** unlaudable has joined #openstack-ansible21:24
mhaydenApsu: i updated that commit21:25
*** cemmason has quit IRC21:25
odyssey4memhayden feedback on https://review.openstack.org/287440 - I'm not sure that I'm right, but suspect that I am :)21:26
odyssey4meyou should possibly test21:26
mhaydenyou might be right21:27
mhaydenyour method is a bit cleaner21:27
odyssey4memhayden another comment21:27
odyssey4mesudden realisation :)21:27
mhaydeni see your point, but i might check for it anyway21:27
mhaydenjust in case someone removed it21:28
mhayden:P21:28
*** unlaudable has quit IRC21:28
odyssey4memhayden then a lot more would blow up21:28
mhaydentrue21:28
mhaydenbut my task wouldn't!21:28
odyssey4methe template conditionals expect it to be defined21:28
* mhayden chuckles21:28
* mhayden serious'd21:28
odyssey4me:p21:28
mhaydenodyssey4me: relevant -> https://s-media-cache-ak0.pinimg.com/736x/c4/21/3e/c4213e6c108dbf313801958d4cd67b16.jpg21:29
*** unlaudable has joined #openstack-ansible21:29
odyssey4mehaha21:29
odyssey4meanyone for Rocket League on XBox?21:29
odyssey4meit's time to stop computering21:29
palendaeodyssey4me: pft, xbox21:30
openstackgerritMajor Hayden proposed openstack/openstack-ansible-os_keystone: Remove dangling Default domain cfg file  https://review.openstack.org/28744021:30
palendaeAlso Rocket League is like typing with oven mitts :p21:30
mhaydenXBOX? more like http://i.imgur.com/TlqWzC8.jpg21:31
* mhayden waits for redditors to come to my rescue21:31
mhaydenah, i messed up pronouns -- i guess that makes it coffee time21:32
openstackgerritNolan Brubaker proposed openstack/openstack-ansible: Update user variables to Liberty names  https://review.openstack.org/28743821:33
*** jguy_ has quit IRC21:34
*** unlaudable has quit IRC21:35
jwitkohey, has anyone every received this error before during the pip install of keystone to the keystone container ?21:38
jwitkoCollecting testtools<2.0.0,>=0.9.22 (from fixtures<1.3.0,>=0.3.14->oslo.concurrency<1.9.0,>=1.8.2->keystone)21:38
jwitko  Could not find a version that satisfies the requirement testtools<2.0.0,>=0.9.22 (from fixtures<1.3.0,>=0.3.14->oslo.concurrency<1.9.0,>=1.8.2->keystone) (from versions: 2.0.0)21:38
jwitkoNo matching distribution found for testtools<2.0.0,>=0.9.22 (from fixtures<1.3.0,>=0.3.14->oslo.concurrency<1.9.0,>=1.8.2->keystone)21:39
jwitkoAttachment21:39
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [WIP] Gate: Set pip index and extra-index to OpenStack-CI repositories  https://review.openstack.org/26586721:39
*** dnt has quit IRC21:41
*** willlo has joined #openstack-ansible21:41
mhaydenpalendae: i'm reviewing https://review.openstack.org/#/c/286753/11/scripts/upgrade-utilities/scripts/add_new_neutron_env.py -- what's the flag file?21:41
jwitkoit looks like my repo is missing a package ?21:41
jwitkothe repo was just built in the previous steps though.  and i built it locally, not with mirrors21:41
palendaemhayden: In the same review, https://review.openstack.org/#/c/286753/11/doc/source/upgrade-guide/overview.rst21:41
mhaydenpalendae: ah, just to keep track of what's been moved from kilo > liberty so far? that makes sense21:42
palendaemhayden: And the creates line in https://review.openstack.org/#/c/286753/11/doc/source/upgrade-guide/overview.rst21:42
palendaeEr, https://review.openstack.org/#/c/286753/11/scripts/upgrade-utilities/playbooks/deploy-config-changes.yml21:42
palendaemhayden: It's so that the calling Ansible task doesn't re-run21:42
mhaydengot it21:42
*** johnmilton has quit IRC21:42
palendaeThe scripts themselves *should* be harmless21:43
mhaydenthe logic makes sense now -- thakns21:43
palendaeIf run 2+ times21:43
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [WIP] Gate: Use apt sources configured on the host  https://review.openstack.org/27963321:44
odyssey4mejwitko did you execute repo-build? which branch/tag?21:45
jwitkoodyssey4me, yes I did.  keystone.21:45
jwitkoa lot of other packages make it in just fine21:45
odyssey4mejwitko which branch/tag are you building from?21:45
odyssey4mejwitko also, have you verified that the repo servers all have the same content?21:45
jwitkohttp://10.96.112.34:8181/os-releases/11.2.9/21:45
automagicallyLater all21:45
jwitkoodyssey4me, oh is it possible they're not sync'd yet ?21:46
odyssey4mejwitko it is21:46
odyssey4meit's also possible that wheel has built wheels that pip can't install21:47
odyssey4mejwitko https://github.com/openstack/openstack-ansible/commit/81f45653fa6bba4bc63fc44785b422b44f321a1721:47
jwitkoodyssey4me, so i see the package21:47
odyssey4meif you're building your own repo then that's a possible cause21:47
jwitkohttp://10.96.112.34:8181/os-releases/11.2.9/testtools-2.0.0-py2.py3-none-any.whl21:47
jwitkothe requirements are for <2.0.0 as per the error message above21:47
jwitkoand the package built into 11.2.9 repo seems to be 2.0.0 ?21:48
*** unlaudable has joined #openstack-ansible21:48
Apsumhayden: +121:48
odyssey4mejwitko are you using your own mirror/fork of the git repositories?21:48
jwitkoyes, i built my own21:49
odyssey4mejwitko do your git url's have '.git' at the end of them?21:49
jwitkosorry, i don't know what you mean by git urls ?21:49
jwitkowhere do I see htose?21:49
jwitkothose*?21:49
*** b3rnard0 is now known as b3rnard0_away21:50
odyssey4mejwitko eg: https://github.com/openstack/openstack-ansible/blob/kilo/playbooks/defaults/repo_packages/openstack_services.yml#L3321:50
jwitko## Global Requirements21:51
jwitkorequirements_git_repo: https://git.openstack.org/openstack/requirements21:51
jwitkorequirements_git_install_branch: 0517298926fa413c3aa03d7e93d5a21bdb9d6ca9 # HEAD of "stable/kilo" as of 22.01.201621:51
jwitkorequirements_git_dest: "/opt/requirements_{{ requirements_git_install_branch | replace('/', '_') }}"21:51
odyssey4mejwitko ok, so you haven't changed those vars to a local git repo21:51
jwitkoand I need to?  wouldn't the head of stable/kilo be fine ?21:51
openstackgerritNate Potter proposed openstack/openstack-ansible: Add haproxy-install to setup-infrastructure.yml  https://review.openstack.org/28746821:52
odyssey4mejwitko you should have a report that looks like this - can you pastebin it? http://rpc-repo.rackspace.com/reports/11.2.9.json21:52
odyssey4mejwitko you don't need to - that's just something that sometimes causes issue like you're experiencing - I'm just eliminating possible root causes21:53
jwitkoodyssey4me, http://paste.openstack.org/raw/489058/21:53
admin0guys .. new setup, new install .. what could cause this ? http://pastebin.com/2vY83q0J  - KeyError: 'children'21:58
jwitkochildren are subsets of host groups,  check your keystone hosts in your user config?21:59
*** thorst has quit IRC22:01
odyssey4mejwitko ok, that looks fine - are you able to share your useR_variables file?22:02
jwitkosure, any specific part?22:02
*** thorst has joined #openstack-ansible22:02
odyssey4mejwitko anything that's not a default22:02
openstackgerritNate Potter proposed openstack/openstack-ansible: Clarify HAProxy configuration necessity  https://review.openstack.org/28747122:04
jwitkoodyssey4me, here is the entire file minus comments:  http://paste.openstack.org/raw/lFlpChevBV5FKLTmG9Bw/22:04
jwitkonothing specific to repo22:04
openstackgerritBjoern Teipel proposed openstack/openstack-ansible: Adding release notes for erlang VM tuning parameters  https://review.openstack.org/28550622:05
*** neilus has quit IRC22:06
*** thorst has quit IRC22:06
openstackgerritJesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [WIP] Only include common overrides in user_variables.yml  https://review.openstack.org/28747422:08
*** mathlin1 has quit IRC22:09
*** albertcard has quit IRC22:09
*** woodard has quit IRC22:09
jwitkoalso, odyssey4me, i went into a repo container and checked the wheel version22:10
jwitkoit is 0.26.022:10
jwitkowhich is the version that fix you linked me to would set it at22:11
jwitkoso i think we're good on that end22:11
odyssey4mejwitko ok - I need to stop computering and can't really think much any more (it's 10:15pm)... hopefully someone else can help you get to the bottom of it22:11
jwitkooh sorry didn't realize your time different.  thanks for your help, have a good night22:12
palendaeI'm off for a while, so no changes to https://review.openstack.org/#/q/topic:liberty_upgrade if people are looking at them. I'll probably check in to poke the gates later if it's necessary, though22:15
*** sdake_ has joined #openstack-ansible22:15
*** sdake has quit IRC22:18
*** Mudpuppy has quit IRC22:18
*** spotz is now known as spotz_zzz22:18
*** weshay has quit IRC22:19
admin0the deploy hsot need to be able to reach the management network because its using the management network for containers ssh also right ?22:19
*** b3rnard0_away is now known as b3rnard022:20
jwitkoadmin0, yes correct22:20
admin0ok22:20
jwitkoor the deploy host can be inside of the management network22:21
*** thorst has joined #openstack-ansible22:23
*** thorst has quit IRC22:27
admin0how many times for  different cloud installations ( rough figures ) is this ansible playbook used inside rackspace ?22:29
admin0i need to sell this to mgmt tomorrow as why i am a big fan of this and recommend this as the next method22:29
admin01000+ ?22:30
*** elopez has quit IRC22:31
bsvadmin0: being able to reproduce your entire setup with a playbook, should be reason enough :)22:35
bsvdisaster recovery +++22:36
admin0:D22:36
bsvcompliance ++++22:36
bsvdocumentation ++22:36
bsv;)22:36
admin0it must have taken 1000s of tries to get this playook right22:36
*** neilus has joined #openstack-ansible22:37
admin0how do I force-redo a container creation ?22:44
bgmccollumjust a single container?22:44
admin0something happened . on the start of setup-hosts, it says c11_xxx_container failed22:44
admin0and its not even trying to create it22:45
bgmccollumyou can `lxc-stop -n $CONTAINER_NAME` then `lxc-destroy -n $CONTAINER_NAME`...then rerun playbooks and it will get recreated22:45
admin0yes, but that container is not there in the first place22:46
admin0it says22:46
admin0SSH Error: data could not be sent to the remote host. Make sure this host can be reached over ssh22:46
bgmccollumo_O22:46
admin0so i think it assumes its there,22:46
admin0and not tries to recreate22:46
bgmccollumwhich task?22:46
admin0lxc-ls is blank22:46
bsvi wasted 3-4 hours today, running playbooks, mysql-container failed all the time... until we discovered that someone had not run the ha-proxy playbook >.<22:46
admin0setup-hosts22:46
bgmccollumright, but which task inside that playbook?22:47
admin0bgmccollum: http://pastebin.com/N4HVWbDJ22:48
admin0its trying to gather facts of a non-existent container22:48
bgmccollumwell, it might be related to the `--limit=c11_galera_container-f0a631d3` in your command...22:49
admin0i did that iwthout it also .. same22:49
admin0somehow its skipping assuming a container that is not even there in the first place22:49
admin0maybe delete from inventory and it might try to create again22:49
bgmccollumadmin0 how vanilla is this setup?22:50
admin0brand new22:50
bgmccollumi mean, you're not doing any customer env.d stuff, right?22:50
bgmccollumcustom*22:50
admin0i will redo22:50
*** yarkot_ has joined #openstack-ansible22:53
bgmccollumadmin0 did you try to move galera to `is_metal: true` or something?22:56
admin0i removed the is_metal22:56
bgmccollumah...22:56
admin0figured out except neutro and haproxy, everything else is good in containers :D)22:57
*** yarkot_ has quit IRC22:57
*** galstrom is now known as galstrom_zzz22:57
*** busterswt has quit IRC23:00
*** eil397 has quit IRC23:01
admin0everythign back on track again :D23:07
admin0there was one ip mapping file23:07
admin0that was causing this23:07
*** izaakk has quit IRC23:07
admin0maybe from a previous run control C in the middle stuff23:07
* admin0 notes to self: never do a control C in mid, change user_file and expect it to be flawless :D 23:09
*** dnt has joined #openstack-ansible23:12
*** dnt has left #openstack-ansible23:12
*** retreved has quit IRC23:13
*** sigmavirus24 is now known as sigmavirus24_awa23:15
*** albertcard has joined #openstack-ansible23:17
jwitkohey bgmccollum, can you help me with a repo issue?  I built a local repo during a fresh kilo install with openstack-ansible repo release 11.2.9,  and I'm seeing the following error when trying to do the pip install of keystone onto the keystone container.  There are tons of other keystone related pip libraries that install before without any issue.  Seems to be a very mismatch problem23:24
jwitkovia a dependency?   http://paste.openstack.org/raw/zTlWg2aZY32xSb1krJx7/23:24
jwitkoor cloudnull, if you're around23:24
bgmccollumjwitko, let me kick off a fresh kilo install and see what happens. are you using a tag, or the head of the kilo branch?23:27
jwitkobgmccollum, i'm using the 11.2.9 release tag23:28
bgmccollumthanks23:28
*** KLevenstein has quit IRC23:30
jwitkobgmccollum, earlier odyssey4me pointed out https://github.com/openstack/openstack-ansible/commit/81f45653fa6bba4bc63fc44785b422b44f321a1723:31
jwitkobut it doesn't look like I'm having that issue23:32
jwitkoroot@oss-ctrl02_keystone_container-259c49f7:~# pip list  | grep wheel23:32
jwitkowheel (0.26.0)23:32
bgmccollumjwitko can you look at the version of wheel in your repo container?23:32
jwitkoroot@oss-ctrl02_repo_container-cba2bb4c:~# pip list | grep wheel23:33
jwitkowheel (0.26.0)23:33
*** johnmilton has joined #openstack-ansible23:35
*** subscope has quit IRC23:37
*** admin0 has quit IRC23:38
bgmccollumjwitko ill let you know what happens...23:38
jwitkothanks23:38
*** jmckind has joined #openstack-ansible23:41
*** jmckind has quit IRC23:42
*** jguy_ has joined #openstack-ansible23:42
*** jmckind has joined #openstack-ansible23:42
jwitkobgmccollum, it looks like the error lies in the fixtures package,  the inheritance is "keystone --> oslo.concurrency --> fixtures23:50
jwitkohey bgmccollum, i'm looking at the repo for rpc-repo.rackspace.com,  http://rpc-repo.rackspace.com/os-releases/11.2.9/23:53
jwitkoit has testtools=1.9.023:53
jwitkowhy is my repo different?23:53
jwitkodid I just clone it at a bad time ?23:53
bgmccollumjwitko did you run the clone, or build playbook?23:53
*** kencjohnston has quit IRC23:53
jwitkobuild23:53
*** jmckind is now known as jmckind_23:54
bgmccollumi think build might result it something different, as uncapped requirement might drift into breakage as you've discovered. might need to pin something new. if you do repo clone, you will get exactly whats in the RPC repo.23:55
bgmccollumi didnt know you were building your own repo instead of cloning, so my build is likely to work.23:55
jwitkobgmccollum, so do you have a suggested course of action ?23:56
jwitkocan I just replace the rpm ?23:56
jwitkoerr, lol.  .whl23:56
bgmccollumclone the RPC repo instead of building from scratch23:56
bgmccollumthere are likely other nuanced differences in versions23:56
jwitkoand so in repo-install.yml i should replace "repo-build.yml" with "repo-clone-mirror.yml" ?23:56

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!