*** flemingo has quit IRC | 00:00 | |
*** sawblade_ has joined #openstack-ansible | 00:07 | |
*** flemingo has joined #openstack-ansible | 00:08 | |
*** sawblade6 has quit IRC | 00:10 | |
*** flemingo has quit IRC | 00:12 | |
*** odyssey4me has quit IRC | 00:14 | |
*** odyssey4me has joined #openstack-ansible | 00:14 | |
*** markvoelker has quit IRC | 00:21 | |
*** flemingo has joined #openstack-ansible | 00:29 | |
*** flemingo has quit IRC | 00:33 | |
*** flemingo has joined #openstack-ansible | 00:54 | |
*** sawblade6 has joined #openstack-ansible | 00:57 | |
*** flemingo has quit IRC | 00:59 | |
*** sawblade_ has quit IRC | 01:00 | |
*** flemingo has joined #openstack-ansible | 01:07 | |
idlemind | cloudnull your patches to use systemd for network config ... i'm all for. it looks like we're having to use a custom systemd service for "post-up" type of actions though. is that a limitation we expect to go away. reading up on the topic on arch-wiki it seems to be a component missing from the spec atm | 01:10 |
---|---|---|
*** flemingo has quit IRC | 01:11 | |
*** fatdragon has joined #openstack-ansible | 01:18 | |
*** markvoelker has joined #openstack-ansible | 01:19 | |
*** fatdragon has quit IRC | 01:23 | |
*** markvoelker has quit IRC | 01:52 | |
*** fatdragon has joined #openstack-ansible | 01:55 | |
*** fatdragon has quit IRC | 02:06 | |
*** flemingo has joined #openstack-ansible | 02:07 | |
*** flemingo has quit IRC | 02:12 | |
*** fatdragon1 has joined #openstack-ansible | 02:18 | |
*** gkadam__ has joined #openstack-ansible | 02:18 | |
*** flemingo has joined #openstack-ansible | 02:22 | |
*** flemingo has quit IRC | 02:27 | |
*** markvoelker has joined #openstack-ansible | 02:49 | |
*** flemingo has joined #openstack-ansible | 02:51 | |
*** admin0 has joined #openstack-ansible | 02:52 | |
*** flemingo has quit IRC | 02:56 | |
*** admin0 has quit IRC | 02:56 | |
*** mpjetta has joined #openstack-ansible | 02:58 | |
*** flemingo has joined #openstack-ansible | 03:10 | |
*** flemingo has quit IRC | 03:14 | |
*** markvoelker has quit IRC | 03:23 | |
*** flemingo has joined #openstack-ansible | 03:33 | |
*** flemingo has quit IRC | 03:37 | |
cloudnull | idlemind: yea we're using a oneshot service for post up scripts. it seems to be a more universal way to accomplish the same goal as the debian post-up scripts. | 03:39 |
cloudnull | but if there's something else we can do, thats better suited, I'm game . | 03:39 |
*** sawblade_ has joined #openstack-ansible | 03:56 | |
*** sawblade6 has quit IRC | 03:58 | |
*** fatdragon has joined #openstack-ansible | 04:03 | |
openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: [TEST] Update Ansible to 2.5.0rc2 https://review.openstack.org/501814 | 04:09 |
*** sawblade6 has joined #openstack-ansible | 04:11 | |
openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Adding openstack_service_venv_bin else symlink creation in ceph_client role fails. https://review.openstack.org/488976 | 04:11 |
openstackgerrit | Merged openstack/openstack-ansible stable/queens: [Docs] Include OpenStack-Ansible Manifesto https://review.openstack.org/556362 | 04:13 |
*** fatdragon has quit IRC | 04:14 | |
*** sawblade_ has quit IRC | 04:14 | |
openstackgerrit | Merged openstack/openstack-ansible stable/queens: [Docs] Restructure inventory documentation https://review.openstack.org/556378 | 04:14 |
*** armaan has quit IRC | 04:14 | |
*** armaan has joined #openstack-ansible | 04:15 | |
*** markvoelker has joined #openstack-ansible | 04:20 | |
*** fatdragon1 has quit IRC | 04:21 | |
*** sawblade_ has joined #openstack-ansible | 04:30 | |
*** sawblade6 has quit IRC | 04:30 | |
openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: [DNM] Testing tempest with longer timeouts https://review.openstack.org/558098 | 04:31 |
*** gkadam__ has quit IRC | 04:33 | |
idlemind | cloudnull ya i don't think there's a pre-built networkd way yet so the oneshot service is the only answer ... on centos i switched to networkmanager that has pre/post - up and pre/post - down methods but i know a lot of ppl get jittery about having networkmanager on | 04:44 |
idlemind | and i like the idea of networkd being more likely to be cross platform is more appealing though | 04:45 |
*** flemingo has joined #openstack-ansible | 04:46 | |
*** gkadam has joined #openstack-ansible | 04:47 | |
idlemind | even if that means using our "own" post-up type replacement | 04:49 |
*** flemingo has quit IRC | 04:51 | |
*** markvoelker has quit IRC | 04:53 | |
*** sawblade6 has joined #openstack-ansible | 05:15 | |
*** sawblade_ has quit IRC | 05:18 | |
*** flemingo has joined #openstack-ansible | 05:48 | |
*** markvoelker has joined #openstack-ansible | 05:50 | |
*** flemingo has quit IRC | 05:52 | |
*** sawblade_ has joined #openstack-ansible | 05:59 | |
*** sawblade6 has quit IRC | 06:01 | |
*** flemingo has joined #openstack-ansible | 06:02 | |
*** flemingo has quit IRC | 06:07 | |
*** fatdragon has joined #openstack-ansible | 06:11 | |
*** fatdragon has quit IRC | 06:21 | |
*** flemingo has joined #openstack-ansible | 06:22 | |
*** markvoelker has quit IRC | 06:23 | |
*** flemingo has quit IRC | 06:26 | |
*** sawblade6 has joined #openstack-ansible | 06:27 | |
*** sawblade_ has quit IRC | 06:29 | |
*** sawblade_ has joined #openstack-ansible | 06:36 | |
*** flemingo has joined #openstack-ansible | 06:36 | |
*** sawblade6 has quit IRC | 06:39 | |
*** flemingo has quit IRC | 06:41 | |
*** flemingo has joined #openstack-ansible | 06:49 | |
*** flemingo has quit IRC | 06:53 | |
*** flemingo has joined #openstack-ansible | 07:09 | |
*** flemingo has quit IRC | 07:15 | |
*** admin0 has joined #openstack-ansible | 07:17 | |
*** NWOutlier__ has joined #openstack-ansible | 07:19 | |
*** armaan has quit IRC | 07:36 | |
*** armaan has joined #openstack-ansible | 07:36 | |
*** sawblade6 has joined #openstack-ansible | 07:44 | |
*** sawblade_ has quit IRC | 07:47 | |
*** fatdragon has joined #openstack-ansible | 07:50 | |
*** admin0 has quit IRC | 07:58 | |
*** admin0 has joined #openstack-ansible | 07:58 | |
*** fatdragon has quit IRC | 08:02 | |
*** flemingo has joined #openstack-ansible | 08:03 | |
*** armaan has quit IRC | 08:04 | |
*** armaan has joined #openstack-ansible | 08:05 | |
*** flemingo has quit IRC | 08:08 | |
*** flemingo has joined #openstack-ansible | 08:14 | |
*** armaan has quit IRC | 08:17 | |
*** armaan has joined #openstack-ansible | 08:17 | |
*** flemingo has quit IRC | 08:19 | |
*** markvoelker has joined #openstack-ansible | 08:20 | |
*** sawblade_ has joined #openstack-ansible | 08:25 | |
*** sawblade6 has quit IRC | 08:27 | |
*** gkadam_ has joined #openstack-ansible | 08:28 | |
*** Bico_Fino has quit IRC | 08:30 | |
*** haad1 has joined #openstack-ansible | 08:30 | |
*** gkadam has quit IRC | 08:31 | |
*** Bico_Fino has joined #openstack-ansible | 08:32 | |
*** flemingo has joined #openstack-ansible | 08:35 | |
*** flemingo has quit IRC | 08:41 | |
*** sawblade_ has quit IRC | 08:45 | |
*** sawblade6 has joined #openstack-ansible | 08:51 | |
*** markvoelker has quit IRC | 08:55 | |
*** yolanda__ has joined #openstack-ansible | 08:55 | |
*** yolanda_ has quit IRC | 08:55 | |
*** armaan has quit IRC | 09:16 | |
*** armaan has joined #openstack-ansible | 09:17 | |
*** sawblade6 has quit IRC | 09:19 | |
*** sawblade6 has joined #openstack-ansible | 09:21 | |
_moep_ | hey! whats the OSA package name at debian 9? or is it not possible to use it there? I was reading about e.g. ubuntu | 09:23 |
*** Sha0000 has joined #openstack-ansible | 09:26 | |
*** armaan has quit IRC | 09:27 | |
*** armaan has joined #openstack-ansible | 09:27 | |
*** NWOutlier__ has quit IRC | 09:29 | |
*** sawblade_ has joined #openstack-ansible | 09:40 | |
*** sawblade6 has quit IRC | 09:40 | |
*** markvoelker has joined #openstack-ansible | 09:52 | |
*** sawblade6 has joined #openstack-ansible | 09:58 | |
*** fatdragon has joined #openstack-ansible | 09:59 | |
*** sawblade_ has quit IRC | 10:01 | |
*** sawblade_ has joined #openstack-ansible | 10:04 | |
*** sawblade6 has quit IRC | 10:06 | |
*** fatdragon has quit IRC | 10:09 | |
*** armaan has quit IRC | 10:12 | |
*** armaan has joined #openstack-ansible | 10:13 | |
*** markvoelker has quit IRC | 10:24 | |
*** admin0 has quit IRC | 10:31 | |
*** sawblade6 has joined #openstack-ansible | 10:42 | |
*** sawblade_ has quit IRC | 10:42 | |
*** flemingo has joined #openstack-ansible | 10:45 | |
*** flemingo has quit IRC | 10:50 | |
*** flemingo has joined #openstack-ansible | 10:52 | |
*** flemingo has quit IRC | 10:57 | |
*** sawblade_ has joined #openstack-ansible | 11:04 | |
*** sawblade6 has quit IRC | 11:06 | |
*** flemingo has joined #openstack-ansible | 11:37 | |
*** flemingo has quit IRC | 11:42 | |
*** sawblade6 has joined #openstack-ansible | 11:47 | |
*** sawblade_ has quit IRC | 11:49 | |
*** sawblade_ has joined #openstack-ansible | 11:51 | |
*** sawblade6 has quit IRC | 11:53 | |
*** sawblade6 has joined #openstack-ansible | 12:19 | |
*** sawblade_ has quit IRC | 12:21 | |
*** markvoelker has joined #openstack-ansible | 12:21 | |
*** sawblade_ has joined #openstack-ansible | 12:27 | |
*** sawblade6 has quit IRC | 12:27 | |
*** flemingo has joined #openstack-ansible | 12:28 | |
*** gkadam_ has quit IRC | 12:32 | |
*** flemingo has quit IRC | 12:32 | |
*** flemingo has joined #openstack-ansible | 12:41 | |
*** fatdragon has joined #openstack-ansible | 12:41 | |
*** sawblade6 has joined #openstack-ansible | 12:42 | |
*** odyssey4me has quit IRC | 12:43 | |
*** odyssey4me has joined #openstack-ansible | 12:44 | |
*** sawblade_ has quit IRC | 12:44 | |
*** flemingo has quit IRC | 12:46 | |
*** fatdragon has quit IRC | 12:53 | |
*** markvoelker has quit IRC | 12:55 | |
*** sawblade6 has quit IRC | 12:55 | |
*** sawblade6 has joined #openstack-ansible | 12:56 | |
*** gkadam has joined #openstack-ansible | 12:58 | |
*** sawblade6 has quit IRC | 13:06 | |
*** sawblade6 has joined #openstack-ansible | 13:14 | |
*** fatdragon has joined #openstack-ansible | 13:36 | |
*** flemingo has joined #openstack-ansible | 13:39 | |
*** fatdragon has quit IRC | 13:41 | |
*** flemingo has quit IRC | 13:44 | |
*** Jeffrey4l has quit IRC | 13:48 | |
*** flemingo has joined #openstack-ansible | 13:49 | |
*** Jeffrey4l has joined #openstack-ansible | 13:49 | |
*** flemingo has quit IRC | 13:53 | |
*** haad1 has quit IRC | 13:55 | |
*** sawblade_ has joined #openstack-ansible | 14:04 | |
*** sawblade6 has quit IRC | 14:06 | |
*** fatdragon has joined #openstack-ansible | 14:15 | |
*** fatdragon has quit IRC | 14:20 | |
*** armaan has quit IRC | 14:23 | |
*** armaan has joined #openstack-ansible | 14:23 | |
*** Jeffrey4l has quit IRC | 14:45 | |
*** Jeffrey4l has joined #openstack-ansible | 14:46 | |
*** flemingo has joined #openstack-ansible | 14:48 | |
*** Jeffrey4l has quit IRC | 14:49 | |
*** Jeffrey4l has joined #openstack-ansible | 14:50 | |
*** armaan has quit IRC | 14:51 | |
*** armaan has joined #openstack-ansible | 14:52 | |
*** markvoelker has joined #openstack-ansible | 14:52 | |
*** flemingo has quit IRC | 14:53 | |
*** Jeffrey4l has quit IRC | 14:54 | |
*** Jeffrey4l has joined #openstack-ansible | 14:56 | |
*** Jeffrey4l has quit IRC | 14:56 | |
*** Jeffrey4l has joined #openstack-ansible | 14:59 | |
*** flemingo has joined #openstack-ansible | 15:01 | |
*** flemingo has quit IRC | 15:05 | |
*** sawblade6 has joined #openstack-ansible | 15:08 | |
*** sawblade_ has quit IRC | 15:10 | |
*** sawblade_ has joined #openstack-ansible | 15:20 | |
*** sawblade6 has quit IRC | 15:22 | |
*** markvoelker has quit IRC | 15:26 | |
*** sawblade6 has joined #openstack-ansible | 15:26 | |
*** sawblade_ has quit IRC | 15:28 | |
*** sawblade_ has joined #openstack-ansible | 15:32 | |
*** sawblade6 has quit IRC | 15:34 | |
*** fatdragon has joined #openstack-ansible | 15:44 | |
*** sawblade6 has joined #openstack-ansible | 15:48 | |
*** sawblade_ has quit IRC | 15:50 | |
*** flemingo has joined #openstack-ansible | 15:53 | |
logan- | anyone seeing pip install failures over the past day when an empty package list is provided? | 15:55 |
logan- | admin0 was seeing it on queens yesterday it looks like https://gist.github.com/a1git/b0ab45bd7ba7866b9992782a7cd7c7ab | 15:56 |
logan- | im seeing it on a pike sha that has been pinned and passing since february | 15:56 |
*** flemingo has quit IRC | 15:57 | |
*** armaan has quit IRC | 16:02 | |
logan- | ya also seeing stable/pike failing to bootstrap due to this | 16:02 |
*** armaan has joined #openstack-ansible | 16:02 | |
*** armaan has quit IRC | 16:02 | |
*** armaan has joined #openstack-ansible | 16:03 | |
*** armaan has quit IRC | 16:04 | |
*** armaan has joined #openstack-ansible | 16:04 | |
*** sawblade_ has joined #openstack-ansible | 16:06 | |
*** sawblade6 has quit IRC | 16:09 | |
*** Sha0000 has quit IRC | 16:10 | |
*** flemingo has joined #openstack-ansible | 16:10 | |
*** flemingo has quit IRC | 16:15 | |
*** sawblade6 has joined #openstack-ansible | 16:17 | |
*** sawblade_ has quit IRC | 16:19 | |
logan- | http://paste.ubuntu.com/p/Z2Fp5rkZ9Q/ pike failing | 16:20 |
openstackgerrit | Logan V proposed openstack/openstack-ansible stable/queens: [DNM] Test gate https://review.openstack.org/558116 | 16:21 |
openstackgerrit | Logan V proposed openstack/openstack-ansible stable/pike: [DNM] Test gate https://review.openstack.org/558117 | 16:21 |
*** markvoelker has joined #openstack-ansible | 16:23 | |
openstackgerrit | Logan V proposed openstack/openstack-ansible stable/queens: [DNM] Test gate https://review.openstack.org/558116 | 16:36 |
openstackgerrit | Logan V proposed openstack/openstack-ansible stable/pike: [DNM] Test gate https://review.openstack.org/558117 | 16:36 |
*** Ret64 has joined #openstack-ansible | 16:40 | |
Ret64 | #startmeeting maintenance | 16:40 |
openstack | Meeting started Sun Apr 1 16:40:52 2018 UTC and is due to finish in 60 minutes. The chair is Ret64. Information about MeetBot at http://wiki.debian.org/MeetBot. | 16:40 |
openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 16:40 |
*** openstack changes topic to " (Meeting topic: maintenance)" | 16:40 | |
openstack | The meeting name has been set to 'maintenance' | 16:40 |
*** Ret64 has quit IRC | 16:43 | |
*** armaan has quit IRC | 16:47 | |
*** gkadam has quit IRC | 16:51 | |
*** markvoelker has quit IRC | 16:56 | |
*** flemingo has joined #openstack-ansible | 17:00 | |
*** flemingo has quit IRC | 17:04 | |
*** gkadam has joined #openstack-ansible | 17:11 | |
*** flemingo has joined #openstack-ansible | 17:14 | |
*** flemingo has quit IRC | 17:18 | |
*** Jeffrey4l has quit IRC | 17:25 | |
*** Jeffrey4l has joined #openstack-ansible | 17:25 | |
*** flemingo has joined #openstack-ansible | 17:29 | |
evrardjp | #endmeeting | 17:29 |
evrardjp | oh great | 17:30 |
evrardjp | logan-: pbr patches merged | 17:30 |
logan- | yea doesnt seem related to the pbr stuff | 17:31 |
logan- | showing up in the gate also: https://review.openstack.org/#/c/558116/ (queens) | 17:31 |
logan- | pike: https://review.openstack.org/#/c/558117/ | 17:31 |
evrardjp | damn holidays | 17:33 |
evrardjp | always gate breaking during them | 17:33 |
evrardjp | :p | 17:33 |
evrardjp | (I am checking at pike pip_install role changes) | 17:34 |
*** flemingo has quit IRC | 17:34 | |
evrardjp | spinning up my own vm then | 17:37 |
evrardjp | is it pike/queens only? | 17:37 |
evrardjp | or master/ocata have the same issue? | 17:38 |
logan- | not sure. master does not run the "install pip packages" task during bootstrap-aio so it does not fail immediately like pike/queens | 17:38 |
logan- | ocata i have not tested yet | 17:38 |
evrardjp | let's check zuul data | 17:39 |
logan- | no changes to osa have affected it afaik, like i said im seeing this on a pinned OSA sha from Feb that has been passing reliably until yesterday | 17:39 |
logan- | so im thinking something upstream | 17:39 |
evrardjp | logan-: I am not in the mood for your april's fools joke :p | 17:40 |
evrardjp | hahah | 17:40 |
evrardjp | apparently ocata and newton is quite ok | 17:40 |
logan- | hmm interesting | 17:40 |
evrardjp | master seems to have passed this morning | 17:40 |
*** flemingo has joined #openstack-ansible | 17:42 | |
evrardjp | master upgrades, pike deploys, queens upgrades and queens deploy seem to have failed with that | 17:42 |
evrardjp | that didn't happen in the previous days | 17:43 |
logan- | Successfully installed pip-10.0.0b1 | 17:45 |
logan- | there we go | 17:45 |
logan- | bootstrap hosts runs get-pip.py without any pins | 17:46 |
*** flemingo has quit IRC | 17:47 | |
evrardjp | I am not there yet | 17:48 |
evrardjp | but I compare the logs , and I see some things very weird | 17:48 |
*** sawblade_ has joined #openstack-ansible | 17:50 | |
evrardjp | where do you see that in the logs? | 17:50 |
evrardjp | I'd love if Ret64 would come back and stop the started meeting | 17:51 |
evrardjp | #startmeeting | 17:52 |
openstack | evrardjp: Error: Can't start another meeting, one is in progress. Use #endmeeting first. | 17:52 |
evrardjp | #endmeeting | 17:52 |
*** openstack changes topic to "Launchpad: https://launchpad.net/openstack-ansible || Weekly Meetings: https://wiki.openstack.org/wiki/Meetings/openstack-ansible || Review Dashboard: https://goo.gl/tTmdgs" | 17:52 | |
openstack | Meeting ended Sun Apr 1 17:52:25 2018 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 17:52 |
openstack | Minutes: http://eavesdrop.openstack.org/meetings/maintenance/2018/maintenance.2018-04-01-16.40.html | 17:52 |
openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/maintenance/2018/maintenance.2018-04-01-16.40.txt | 17:52 |
openstack | Log: http://eavesdrop.openstack.org/meetings/maintenance/2018/maintenance.2018-04-01-16.40.log.html | 17:52 |
evrardjp | interesting I couldn't do that earlier | 17:52 |
evrardjp | anyway | 17:52 |
*** sawblade6 has quit IRC | 17:53 | |
evrardjp | logan-: on the failure node: http://logs.openstack.org/periodic/git.openstack.org/openstack/openstack-ansible/stable/pike/openstack-ansible-deploy-aio-ubuntu-xenial/2408ca6/job-output.txt.gz#_2018-04-01_06_09_33_463840 | 17:53 |
logan- | i got that from ara | 17:54 |
evrardjp | mmm | 17:54 |
evrardjp | which job? | 17:54 |
logan- | local job ara | 17:54 |
evrardjp | oh you run that on your machine? | 17:55 |
evrardjp | great | 17:55 |
logan- | no in local gating | 17:55 |
logan- | but im bootstrapping a node atm to confirm | 17:55 |
*** flemingo has joined #openstack-ansible | 17:55 | |
logan- | http://paste.openstack.org/show/718118/ | 17:55 |
logan- | so the ansible-runtime venv has pinned pip, but we install unpinned on the host | 17:56 |
evrardjp | so you think it's the new get-pip that doesn't respect our current arguments? | 17:57 |
evrardjp | that's why we should vendor this script in | 17:57 |
logan- | let me see if we are passing any arguments | 17:57 |
evrardjp | I see | 17:58 |
evrardjp | https://github.com/openstack/openstack-ansible-pip_install/blob/stable/pike/tasks/install_source.yml#L19-L22 | 17:58 |
logan- | and this is why pip 10 is the problem: http://paste.openstack.org/raw/718119/ | 17:58 |
logan- | passing an empty package list on pip <10 is just fine, but on pip 10 it fails the module :/ | 17:59 |
evrardjp | that seems well empty indeed | 17:59 |
*** flemingo has quit IRC | 18:00 | |
evrardjp | yup | 18:00 |
logan- | yea ara output from the get-pip install http://paste.openstack.org/raw/718120/ | 18:00 |
evrardjp | what do you think we always vendor the get-pip in, so that we don't bump the version of pip (get-pip) without our knowledge? | 18:00 |
logan- | we don't need to vendor get-pip | 18:01 |
logan- | we need to pass the pins to it | 18:01 |
logan- | the problem is not that get pip doesn't respect our pins, its that we didn't pass any | 18:01 |
evrardjp | that's another way to do it | 18:01 |
evrardjp | yeah | 18:01 |
logan- | it will always try to install the latest release even if you use an older version of get-pip iirc | 18:01 |
evrardjp | well | 18:01 |
evrardjp | the failed task is failing on: /usr/local/bin/pip2 install -U --constraint /home/zuul/src/git.openstack.org/openstack/openstack-ansible/global-requirement-pins.txt --constraint https://git.openstack.org/cgit/openstack/requirements/plain/upper-constraints.txt?id=6ea2d17f0a90985aa05c809c3e536abb7ea74b60 | 18:02 |
evrardjp | but indeed we are talking about the rest | 18:02 |
logan- | yea the task before that is the one where the problem lies | 18:02 |
logan- | pip_install : Install PIPlocalhost | 18:02 |
evrardjp | my problem is | 18:03 |
evrardjp | if we don't change what we are doing right now | 18:03 |
evrardjp | we'll always have a different version of get pip | 18:03 |
evrardjp | https://bootstrap.pypa.io/get-pip.py | 18:03 |
evrardjp | isn't specified at a point in time | 18:04 |
evrardjp | so yes, we can pass a proper list of packages and proper constraints to get pip script | 18:04 |
evrardjp | but that script itself changes | 18:04 |
evrardjp | or is my understanding wrong? | 18:04 |
logan- | yes but that doesn't matter for the current problem | 18:04 |
*** flemingo has joined #openstack-ansible | 18:05 | |
logan- | the latest get-pip will install whatever version you tell it | 18:05 |
evrardjp | I am not sure I understand what you mean. Do you want to use the global-requirements-pins as a list of packages to install to get-pip, by changing https://github.com/openstack/openstack-ansible/blob/stable/queens/tests/bootstrap-aio.yml#L22 ? | 18:07 |
logan- | let me show you the reason ocata doesn't break | 18:07 |
evrardjp | ok | 18:07 |
evrardjp | I feel dumb now | 18:07 |
logan- | https://github.com/openstack/openstack-ansible/blob/1dc4e1baae67f535780d9ea0b84c2d78eb38e8f2/playbooks/inventory/group_vars/all.yml#L52-L57 | 18:07 |
logan- | for some reason this is not set in pike or queens anymore | 18:08 |
logan- | the release bump script used to maintain those pins along with global-requirement-pins.txt | 18:08 |
*** flemingo has quit IRC | 18:09 | |
evrardjp | yes that's true, I don't bump them anymore | 18:09 |
logan- | ya but not setting pip_packages means we get unpinned pip in all of our hosts | 18:10 |
evrardjp | yup | 18:10 |
logan- | well maybe not, i guess we get pinned pip after repo is built since global-requirement-pins should be used there | 18:11 |
evrardjp | so let's think about what we should do | 18:11 |
logan- | why do we bother maintaining pip pins in global-requirements-pins if we arent going to pin it in pip_packages also | 18:12 |
evrardjp | not sure what you meant there, but I agree with you with the one source of truth there. | 18:12 |
evrardjp | So | 18:12 |
evrardjp | I am not sure where the problem lies | 18:12 |
evrardjp | For gates, I am sure we have a problem because the pip_install in the boostrap doesn't get that list | 18:13 |
evrardjp | so that host is bad | 18:13 |
evrardjp | and later in the process, if you have multiple nodes, we do kind a the same with the pip install called in the openstack_hosts role | 18:13 |
evrardjp | iirc | 18:13 |
evrardjp | (talking about master) | 18:14 |
evrardjp | and queens | 18:14 |
evrardjp | but if we take a step back, should we care? | 18:14 |
evrardjp | in emergency, we can indeed pass the global-requirements-pins as a requirement for the boostrap host | 18:14 |
evrardjp | but that won't find the rest of the hosts | 18:14 |
evrardjp | using defaults in the pip_install role seems okay | 18:15 |
evrardjp | so I mean changing that https://github.com/openstack/openstack-ansible-pip_install/blob/stable/ocata/defaults/main.yml#L87-L90 | 18:15 |
evrardjp | but I am not sure what you're proposing | 18:16 |
logan- | well we used to care a lot about pinning pip, and we did that by maintaining global-requirement-pins.txt, and pip_packages in group_vars/all | 18:19 |
logan- | at some point, either by accident or on purpose, we stopped pinning in group_vars/all, but kept doing it in global-requirement-pins, which means we now only partially pin pip | 18:20 |
logan- | personally i think time has proven we need to completely pin pip and setuptools over and over again | 18:21 |
evrardjp | logan-: when reading this: https://github.com/openstack/openstack-ansible/commit/5f7000e185fca71a5077ae7e2bcd29450c4cb70f#diff-9893e2ae9f6cad034565d01efeaeba92 | 18:22 |
evrardjp | the intent is clearly letting pip_install do the right thing on the hosts | 18:22 |
evrardjp | logan-: I think it's fine if we put a pin inside the pip_install role, in the defaults | 18:23 |
evrardjp | because that ENSURES on all hosts a certain version is set | 18:23 |
evrardjp | that's our standard interface | 18:23 |
evrardjp | not sure if you agree | 18:24 |
evrardjp | let's evaluate if global-requirement-pins have another use later | 18:24 |
*** flemingo has joined #openstack-ansible | 18:24 | |
logan- | yea i can think of reasons to do it in pip_install vs integrated repo, not sure what is the optimal choice for the location of the pins. i just think they need to exist somewhere (whether its pip_install defaults or integrated repo vars) | 18:25 |
evrardjp | what I said with the fact of vendoring/pinning get-pip is just because I foresee another issue there. | 18:25 |
evrardjp | logan-: well the pip_install role already lists them as things to install on the host | 18:26 |
evrardjp | https://github.com/openstack/openstack-ansible-pip_install/blob/stable/queens/defaults/main.yml#L49-L51 | 18:26 |
evrardjp | why not adding a version | 18:26 |
evrardjp | :p | 18:26 |
logan- | well unless we stop pinning those packages in global-requirement-pins somehow, then it will make your release management more challenging, because you'll need to maintain matching pins between both repos anytime you want to cut a release | 18:27 |
evrardjp | my concern is that passing the global-requirements-pins as an extra constraint for the bootstrap-aio, like we do everywhere else, would just work on the deploy node | 18:27 |
logan- | im guessing that is one reason we maintained the pins all in one place (integrated repo) before | 18:27 |
evrardjp | so we'd sync anyway for the other nodes | 18:28 |
evrardjp | logan-: the current state of the affairs is that I can do them separately | 18:28 |
evrardjp | but my problem is that: | 18:29 |
*** flemingo has quit IRC | 18:29 | |
evrardjp | if we move to use pip_install only (which is fine for me) | 18:29 |
evrardjp | we'll have to think about all the rest of the process: | 18:29 |
evrardjp | http://codesearch.openstack.org/?q=global-requirement-pins&i=nope&files=&repos= | 18:29 |
*** sawblade6 has joined #openstack-ansible | 18:29 | |
evrardjp | global requirement pins was a top constraint that was not only for pip/setuptools/wheels | 18:30 |
evrardjp | there is still something I don't understand | 18:30 |
evrardjp | http://git.openstack.org/cgit/openstack/openstack-ansible/tree/tests/bootstrap-aio.yml#n30 | 18:30 |
evrardjp | oh no my bad | 18:31 |
*** sawblade_ has quit IRC | 18:32 | |
evrardjp | I was thinking of re-using this: pip_install_upper_constraints into the pip_install getpip task, but that won't help on all the other nodes. | 18:33 |
openstackgerrit | Logan V proposed openstack/openstack-ansible-pip_install master: Implement stable package pins https://review.openstack.org/558119 | 18:33 |
*** flemingo has joined #openstack-ansible | 18:34 | |
evrardjp | you can have those in vars! | 18:34 |
evrardjp | :) | 18:34 |
openstackgerrit | Logan V proposed openstack/openstack-ansible master: Use stable pip pins https://review.openstack.org/558120 | 18:34 |
logan- | ah yeah good point | 18:35 |
evrardjp | mmm | 18:35 |
evrardjp | I am not sure | 18:35 |
openstackgerrit | Logan V proposed openstack/openstack-ansible master: Use stable pip pins https://review.openstack.org/558120 | 18:35 |
evrardjp | This is very conservative | 18:35 |
logan- | the versions you mean? | 18:36 |
evrardjp | I am fine with that, because it's fixing the issue and it's very conservative in what pip install is doing | 18:36 |
evrardjp | but | 18:36 |
logan- | it is the only way we can keep our releases stable | 18:36 |
logan- | bbl gotta head out to brunch, but i wanted to get those up as an idea at least | 18:37 |
evrardjp | yeah | 18:37 |
evrardjp | thanks | 18:37 |
*** flemingo has quit IRC | 18:38 | |
evrardjp | I can't release with those issues pending | 18:43 |
evrardjp | I'd like to have the opinions of others, like odyssey4me or cloudnull | 18:43 |
*** markvoelker has joined #openstack-ansible | 18:53 | |
*** flemingo has joined #openstack-ansible | 18:58 | |
*** haad1 has joined #openstack-ansible | 18:59 | |
*** flemingo has quit IRC | 19:04 | |
*** flemingo has joined #openstack-ansible | 19:25 | |
*** markvoelker has quit IRC | 19:27 | |
*** flemingo has quit IRC | 19:31 | |
*** flemingo has joined #openstack-ansible | 19:34 | |
*** haad1 has quit IRC | 19:35 | |
*** Sha0000 has joined #openstack-ansible | 19:37 | |
*** flemingo has quit IRC | 19:39 | |
*** sawblade_ has joined #openstack-ansible | 19:50 | |
*** sawblade6 has quit IRC | 19:53 | |
*** sawblade6 has joined #openstack-ansible | 19:54 | |
*** sawblade_ has quit IRC | 19:56 | |
*** markvoelker has joined #openstack-ansible | 20:24 | |
*** flemingo has joined #openstack-ansible | 20:30 | |
openstackgerrit | Logan V proposed openstack/openstack-ansible-pip_install master: Implement stable package pins https://review.openstack.org/558119 | 20:33 |
*** flemingo has quit IRC | 20:35 | |
*** haad1 has joined #openstack-ansible | 20:45 | |
openstackgerrit | Logan V proposed openstack/openstack-ansible stable/queens: [DNM] Test gate https://review.openstack.org/558116 | 20:55 |
*** markvoelker has quit IRC | 20:56 | |
*** flemingo has joined #openstack-ansible | 20:58 | |
*** flemingo has quit IRC | 21:03 | |
*** flemingo has joined #openstack-ansible | 21:13 | |
*** flemingo has quit IRC | 21:17 | |
openstackgerrit | Logan V proposed openstack/openstack-ansible-pip_install master: Use constraints with get-pip.py https://review.openstack.org/558124 | 21:21 |
*** flemingo has joined #openstack-ansible | 21:22 | |
logan- | heh so this is the pr that broke our stuff. https://github.com/pypa/pip/pull/4210 | 21:22 |
*** flemingo has quit IRC | 21:26 | |
openstackgerrit | Logan V proposed openstack/openstack-ansible-pip_install stable/queens: Only attempt package installs when packages are provided https://review.openstack.org/558126 | 21:31 |
openstackgerrit | Logan V proposed openstack/openstack-ansible-pip_install stable/queens: Use constraints with get-pip.py https://review.openstack.org/558127 | 21:38 |
logan- | odyssey4me: can you tell me if i'm on the right track here https://review.openstack.org/#/q/topic:get-pip-constraints+OR+topic:skip-package-installs | 21:39 |
openstackgerrit | Logan V proposed openstack/openstack-ansible-pip_install stable/pike: Only attempt package installs when packages are provided https://review.openstack.org/558128 | 21:41 |
*** sawblade6 has quit IRC | 21:42 | |
*** flemingo has joined #openstack-ansible | 21:42 | |
*** sawblade6 has joined #openstack-ansible | 21:43 | |
*** flemingo has quit IRC | 21:47 | |
*** markvoelker has joined #openstack-ansible | 21:53 | |
*** flemingo has joined #openstack-ansible | 21:54 | |
*** sawblade_ has joined #openstack-ansible | 21:56 | |
*** sawblade6 has quit IRC | 21:56 | |
*** flemingo has quit IRC | 21:58 | |
*** sawblade6 has joined #openstack-ansible | 22:04 | |
*** sawblade_ has quit IRC | 22:06 | |
*** ibmko has quit IRC | 22:21 | |
*** sawblade_ has joined #openstack-ansible | 22:25 | |
*** markvoelker has quit IRC | 22:26 | |
*** sawblade6 has quit IRC | 22:27 | |
*** sawblade6 has joined #openstack-ansible | 22:33 | |
*** sawblade_ has quit IRC | 22:36 | |
*** flemingo has joined #openstack-ansible | 22:37 | |
*** flemingo has quit IRC | 22:42 | |
evrardjp | logan-: thanks for the help there | 22:58 |
evrardjp | if you don't mind I'll postpone the release until this gets in | 22:58 |
logan- | cool makes sense | 23:01 |
logan- | you dont have much choice because the pike/queens gates wont pass until we fix it :D | 23:03 |
*** sawblade_ has joined #openstack-ansible | 23:10 | |
*** sawblade6 has quit IRC | 23:12 | |
*** gkadam has quit IRC | 23:17 | |
*** nurdie has quit IRC | 23:20 | |
*** nurdie has joined #openstack-ansible | 23:21 | |
*** flemingo has joined #openstack-ansible | 23:28 | |
logan- | looks like https://review.openstack.org/#/c/558127/ gets queens working again :) | 23:30 |
*** flemingo has quit IRC | 23:32 | |
*** haad1 has quit IRC | 23:33 | |
*** sawblade6 has joined #openstack-ansible | 23:36 | |
*** sawblade_ has quit IRC | 23:39 | |
*** flemingo has joined #openstack-ansible | 23:41 | |
mattt | logan-: how does that bug manifest itself? | 23:45 |
mattt | we have a bunch of broken gates and i suspect it's related | 23:45 |
*** sawblade_ has joined #openstack-ansible | 23:45 | |
*** sawblade6 has quit IRC | 23:45 | |
logan- | pip released a 10.x beta into pypi | 23:46 |
logan- | and we aren't pinning pip installs on pike and above | 23:46 |
*** flemingo has quit IRC | 23:46 | |
logan- | so check pip --version on your failed hosts | 23:46 |
logan- | they probably have pip 10.0.0b1 | 23:47 |
logan- | there is a simple fix here which makes OSA work with the pip 10.x beta: https://review.openstack.org/#/c/558126/ | 23:47 |
logan- | and a more complete fix here that forces osa to actually install the version of pip it is supposed to be pinned to https://review.openstack.org/#/c/558124/ | 23:48 |
mattt | this all affects pike too right logan- ? | 23:49 |
logan- | yes | 23:49 |
logan- | ocata and (probably) newton should be ok because we pinned pip differently there | 23:49 |
mattt | yeah not seeing this on newton | 23:50 |
mattt | ok thanks logan- ! | 23:51 |
logan- | np | 23:51 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!