| SpamapS | zomg I want to deploy the UI with xterm.js like yesterday | 00:01 |
|---|---|---|
| mordred | SpamapS: it's so sexy isn't it? | 00:01 |
| SpamapS | when I was doing my half-arsed ANSI thing I was thinking "isn't this like.. solved?" | 00:02 |
| SpamapS | answer: yes | 00:03 |
| *** rlandy has quit IRC | 00:28 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: webtrigger: add support for parameterized trigger https://review.openstack.org/644484 | 00:58 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add build modal with a parameter form https://review.openstack.org/644485 | 00:58 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add support for checkbox and list parameters https://review.openstack.org/648661 | 00:58 |
| *** swest has quit IRC | 01:39 | |
| *** swest has joined #zuul | 01:55 | |
| *** bhavikdbavishi has joined #zuul | 02:19 | |
| *** bhavikdbavishi has quit IRC | 02:32 | |
| *** bhavikdbavishi has joined #zuul | 03:02 | |
| *** bhavikdbavishi1 has joined #zuul | 03:02 | |
| *** bhavikdbavishi has quit IRC | 03:06 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:06 | |
| *** jamesmcarthur has joined #zuul | 03:18 | |
| tristanC | mordred: just added rdms management to the zuul-operator poc using https://operatorhub.io/operator/alpha/postgres-operator.v3.5.0 | 03:33 |
| *** jamesmcarthur has quit IRC | 03:52 | |
| openstackgerrit | Merged openstack-infra/zuul master: Whitelist pageup and down keys in xterm https://review.openstack.org/649694 | 03:55 |
| tristanC | fwiw, here is the current usage: https://github.com/TristanCacqueray/zuul-operator#usage | 04:00 |
| *** swest has quit IRC | 04:32 | |
| *** spsurya has joined #zuul | 04:32 | |
| *** bjackman has joined #zuul | 04:54 | |
| *** jamesmcarthur has joined #zuul | 05:01 | |
| *** jamesmcarthur has quit IRC | 05:05 | |
| *** pcaruana has joined #zuul | 05:08 | |
| *** swest has joined #zuul | 05:18 | |
| *** swest has quit IRC | 05:22 | |
| *** swest has joined #zuul | 05:38 | |
| *** quiquell|off is now known as quiquell|rover | 05:48 | |
| openstackgerrit | Felix Schmidt proposed openstack-infra/zuul master: Provide roles per tenant via Zuul API https://review.openstack.org/650012 | 06:05 |
| *** bhavikdbavishi has quit IRC | 06:09 | |
| *** bhavikdbavishi1 has joined #zuul | 06:09 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 06:12 | |
| *** quiquell|rover is now known as quique|rover|brb | 06:24 | |
| *** gtema has joined #zuul | 06:51 | |
| *** quique|rover|brb is now known as quiquell|rover | 07:05 | |
| *** gouthamr has quit IRC | 07:11 | |
| *** dmellado has quit IRC | 07:12 | |
| *** dmellado has joined #zuul | 07:18 | |
| *** gouthamr has joined #zuul | 07:20 | |
| *** jpena|off is now known as jpena | 07:52 | |
| *** hashar has joined #zuul | 08:11 | |
| *** jangutter has joined #zuul | 08:12 | |
| jangutter | Hi, I found some weird and unfortunate side-effects with https://review.openstack.org/#/c/609829/ (Cleanup down ports) on our CI. | 08:14 |
| jangutter | We don't run the networking-baremetal ML2 plug-in so baremetal ports are _always DOWN_ (you can't check their state in isolation). | 08:14 |
| jangutter | Mind if I propose configurable knobs to turn off this behaviour? | 08:15 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: Dockerfile: add support for arbritary uid https://review.openstack.org/650246 | 08:35 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add Authorization Rules configuration https://review.openstack.org/639855 | 08:56 |
| *** electrofelix has joined #zuul | 08:58 | |
| *** quiquell|rover is now known as quique|rover|bbl | 09:13 | |
| *** jesusaur has quit IRC | 09:21 | |
| *** jesusaur has joined #zuul | 09:28 | |
| *** hashar has quit IRC | 09:49 | |
| tristanC | jangutter: i think that'll make sense, there is already a 'clean-floating-ips' openstack provider toggle, so I guess there could also be a 'clean-network-ports' provider option | 10:05 |
| *** yolanda has joined #zuul | 10:07 | |
| jangutter | tristanC another option would be to set the "idle timeout" i.e. -1 to never clean, 5 to clean if idle for 5 minutes.... | 10:08 |
| jangutter | tristanC (sean-k-mooney suggested that), will investigate to see if that's hideously complex. | 10:09 |
| *** bhavikdbavishi1 has joined #zuul | 10:14 | |
| *** bhavikdbavishi has quit IRC | 10:17 | |
| *** bhavikdbavishi1 has quit IRC | 10:19 | |
| *** quique|rover|bbl is now known as quiquell|rover | 10:35 | |
| tristanC | mordred: since zuul services scale, i don't think it makes sense to define a fixed quantity for services' instance count... | 10:50 |
| tristanC | mordred: perhaps we should rather define min-instance/max-instance per service instead | 10:50 |
| tristanC | and with the operator reconcile loop, it seems like rather easy to query the gearman bus directly and set the number of replicas accordingly | 10:51 |
| *** gtema has quit IRC | 11:10 | |
| *** EmilienM is now known as EvilienM | 11:14 | |
| *** bhavikdbavishi has joined #zuul | 11:23 | |
| bjackman | When my system comes under very heavy load, I see my jobs failing with "Ansible timeout exceeded: 60" during my pre-run playbooks | 11:36 |
| bjackman | But I don't define job.timeout anywhere | 11:36 |
| bjackman | Any idea why that might be? | 11:36 |
| *** jpena is now known as jpena|lunch | 11:38 | |
| *** yolanda has quit IRC | 11:39 | |
| *** bhavikdbavishi has quit IRC | 11:41 | |
| *** bhavikdbavishi has joined #zuul | 11:45 | |
| *** gtema has joined #zuul | 11:48 | |
| *** quiquell|rover is now known as quique|rover|eat | 11:56 | |
| *** rlandy has joined #zuul | 12:01 | |
| *** quique|rover|eat is now known as quiquell|rover | 12:02 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: web: add tenant and project scoped, JWT-protected actions https://review.openstack.org/576907 | 12:05 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Allow operator to generate auth tokens through the CLI https://review.openstack.org/636197 | 12:05 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Zuul CLI: allow access via REST https://review.openstack.org/636315 | 12:05 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add Authorization Rules configuration https://review.openstack.org/639855 | 12:05 |
| *** hashar has joined #zuul | 12:05 | |
| *** rlandy is now known as rlandy|ruck | 12:10 | |
| *** quiquell|rover is now known as quiquell|lunch | 12:21 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add Authorization Rules configuration https://review.openstack.org/639855 | 12:39 |
| *** jpena|lunch is now known as jpena | 12:40 | |
| *** quiquell|lunch is now known as quiquell|rover | 12:45 | |
| *** jamesmcarthur has joined #zuul | 12:49 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: [WIP] Set focus on xtem.js to terminal https://review.openstack.org/650275 | 12:56 |
| pabelanger | tobiash: any reason not to do^, also I have no idea if that syntax is right, waiting for dashboard job to run | 12:57 |
| *** jamesmcarthur has quit IRC | 13:04 | |
| *** bjackman_ has joined #zuul | 13:05 | |
| *** bjackman has quit IRC | 13:05 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add Authorization Rules configuration https://review.openstack.org/639855 | 13:12 |
| *** jamesmcarthur has joined #zuul | 13:14 | |
| bjackman_ | Do I need to explicitly start the fingergw in my docker-compose.yaml? In my staging setup I haven't done so and I'm able to stream logs but in my prod deployment I'm not able to | 13:16 |
| *** jamesmcarthur has quit IRC | 13:19 | |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: Keep task stdout/stderr separate in result object https://review.openstack.org/650276 | 13:20 |
| pabelanger | bjackman_: you won't be able to change the 60 timeout, that is hard coded. I think we could expose it to an operator, as I too was having the same issue this week. | 13:20 |
| pabelanger | However, my issue was related to lack of entropy for /dev/random | 13:21 |
| pabelanger | bjackman_: shot in the dark, maybe make sure you also have enough: $ cat /proc/sys/kernel/random/entropy_avail | 13:21 |
| bjackman_ | pabelanger, Oh, where is the 60s timeout hard-coded? | 13:21 |
| bjackman_ | Looking at the executor server it seems the relevant timeout comes from the job definition... | 13:22 |
| pabelanger | Oh wait | 13:22 |
| pabelanger | it isn't hardcoded any more | 13:22 |
| bjackman_ | Ahh is it ansible_setup_timeout? | 13:23 |
| pabelanger | https://zuul-ci.org/docs/zuul/admin/components.html#attr-executor.ansible_setup_timeout | 13:23 |
| pabelanger | yah | 13:23 |
| bjackman_ | Ahhh I didn't realise runAnsibleSetup was a separate issue | 13:23 |
| bjackman_ | Awesome, I will try increasing that! | 13:23 |
| bjackman_ | Thanks! | 13:23 |
| pabelanger | however, I would look into your system and try to see why it is taking more then 60. | 13:23 |
| bjackman_ | I believe it's because it was checking out 4 instances of the linux kernel and 8 instances of another repo with over 1000 refs and tags | 13:25 |
| bjackman_ | And it isn't a particularly beefy machine | 13:25 |
| bjackman_ | As for my log streaming issues - I was barking up the wrong tree with fingergw stuff, it seems to be a websockets issue... my web browser (Chrome and Firefox both) sends an HTTP header called "Upgrade" which cherrypy does not like | 13:28 |
| pabelanger | bjackman_: fingergw is only required if you want to use finger cmd and do: $ finger uuid@zuul.example.org | 13:29 |
| bjackman_ | pabelanger, OK thanks | 13:30 |
| *** hashar has quit IRC | 13:32 | |
| mordred | bjackman_: oh weird. we should look in to why the upgrade header isn't working | 13:37 |
| mordred | bjackman_: in production for opendev we have apache running as a proxy in front with mod_proxy_websocket which is maybe doing the upgrade so we're not noticing an issue | 13:37 |
| pabelanger | for nginx, I needed to add the configuration: https://github.com/ansible-network/windmill-config/blob/master/zuul/etc/nginx/sites-available/dashboard-ssl.j2 | 13:39 |
| pabelanger | thanks to SpamapS for config :) | 13:39 |
| *** bjackman_ has quit IRC | 13:48 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: Dockerfile: add support for arbritary uid https://review.openstack.org/650246 | 13:52 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add Authorization Rules configuration https://review.openstack.org/639855 | 14:00 |
| *** johanssone has joined #zuul | 14:03 | |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: Keep task stdout/stderr separate in result object https://review.openstack.org/650276 | 14:04 |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: Fix zuul-stream functional test https://review.openstack.org/650297 | 14:04 |
| *** bjackman_ has joined #zuul | 14:25 | |
| *** bjackman_ has quit IRC | 14:31 | |
| *** rlandy|ruck is now known as rlandy|ruck|mtg | 14:32 | |
| *** jamesmcarthur has joined #zuul | 14:38 | |
| *** quiquell|rover is now known as quiquell|off | 14:47 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add Authorization Rules configuration https://review.openstack.org/639855 | 14:51 |
| *** chandankumar is now known as raukadah | 14:57 | |
| *** hashar has joined #zuul | 14:58 | |
| *** gtema has quit IRC | 14:58 | |
| *** gtema has joined #zuul | 15:00 | |
| *** yolanda has joined #zuul | 15:13 | |
| *** jamesmcarthur_ has joined #zuul | 15:13 | |
| *** rlandy|ruck|mtg is now known as rlandy|ruck | 15:15 | |
| *** jamesmcarthur has quit IRC | 15:16 | |
| *** pwhalen has quit IRC | 15:16 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add Authorization Rules configuration https://review.openstack.org/639855 | 15:22 |
| fbo | Hi, I'm wondering if zuul ops with github projects already seen that issue during a reload https://softwarefactory-project.io/paste/show/1478/ ? | 15:24 |
| pabelanger | tobiash: ^ | 15:27 |
| pabelanger | In a single tenant, do you think we would create a new setting that would allow an operator to limit which pipelines a project could use, when using in-tree zuul.yaml file? I don't think that is possible today, and would require a new tenant with only the specific pipelines right? | 15:40 |
| openstackgerrit | Merged openstack-infra/nodepool master: Use the buildset registry in zuul-quick-start https://review.openstack.org/649468 | 15:40 |
| pabelanger | the use case is, I want to add a project to tenant A, because I haven't created tenant B, but only allow them to add jobs to check pipeline | 15:41 |
| corvus | pabelanger: yeah, i would expect you to just omit the pipelines you don't want used.... | 15:41 |
| corvus | pabelanger: create tenant B? :) | 15:41 |
| pabelanger | okay, that's what I figured, new tenant. Not an issue, was just trying to cut down on some ops work :) | 15:42 |
| *** EvilienM is now known as EmilienM | 15:45 | |
| tobiash | fbo: that's when zuul has no access to the repo | 15:48 |
| tobiash | There is a pr to fix the error message somewhere | 15:48 |
| fbo | tobiash: pabelanger could it be possible a repo has been deleted from the ansible org w/o the removal frm zuul first ? from the log I'm not sur that related to aws. | 15:52 |
| *** yolanda has quit IRC | 15:52 | |
| fbo | tobiash: but also the reconfiguration stop as far as I can see | 15:52 |
| *** jangutter has quit IRC | 16:06 | |
| tobiash | fbo: this will make it tolerate the missing project: https://review.openstack.org/#/c/579872 | 16:11 |
| fbo | tobiash: thanks | 16:13 |
| corvus | pabelanger: ^ you want to +3 that :) | 16:14 |
| *** rlandy|ruck is now known as rlandy|ruck|brb | 16:19 | |
| tobiash | corvus: do we need a third review on 650086 as we both are authors on this? | 16:46 |
| corvus | tobiash: it's a pretty straightforward bugfix, i think it's okay to go through with our +2s | 16:49 |
| tobiash | corvus: k, +3 :) | 16:50 |
| Shrews | fyi, i'm fairly certain the test_handler_poll_session_expired nodepool test is not working at all (thus the constant random failures) | 16:50 |
| pabelanger | sorry, was on public wifi and it died, home now | 16:50 |
| pabelanger | looking | 16:50 |
| *** rlandy|ruck|brb is now known as rlandy|ruck | 16:52 | |
| pabelanger | +3 | 16:52 |
| pabelanger | tobiash: thanks! | 16:52 |
| tobiash | \o/ | 16:55 |
| *** electrofelix has quit IRC | 17:03 | |
| *** jpena is now known as jpena|off | 17:04 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Fix race in test_handler_poll_session_expired https://review.openstack.org/650385 | 17:06 |
| Shrews | i think it was just a race. ^ i think fixes it | 17:06 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Fix race in test_handler_poll_session_expired https://review.openstack.org/650385 | 17:07 |
| corvus | Shrews: suggestion on that ^ | 17:08 |
| Shrews | corvus: *nod* | 17:10 |
| *** gtema has quit IRC | 17:11 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Fix race in test_handler_poll_session_expired https://review.openstack.org/650385 | 17:15 |
| * Shrews wonders what jifflenow is and why he is receiving email to complete my registration there | 17:17 | |
| openstackgerrit | Merged openstack-infra/zuul master: Fix SQL error when querying for tuples https://review.openstack.org/650086 | 17:22 |
| *** jamesmcarthur_ has quit IRC | 17:22 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Fix occasional gate resets with github https://review.openstack.org/650387 | 17:22 |
| *** hashar has quit IRC | 17:22 | |
| tobiash | many words describing a single line fix for sporadic gate resets with github ^ | 17:22 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Fix occasional gate resets with github https://review.openstack.org/650387 | 17:24 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Tolerate missing project https://review.openstack.org/579872 | 17:26 |
| *** jamesmcarthur has joined #zuul | 17:28 | |
| *** manjeets has joined #zuul | 17:32 | |
| *** kmalloc is now known as needscoffee | 17:46 | |
| pabelanger | so, with a promote pipeline, I just noticed the following change A (job running) and change B (waiting), change A finished, when change C entered change queue. But change C started running its job and change B was left waiting. And when change B finally ran, it revered change C, since it didn't have it. Any ideas how to ensure A,B,C run in order? Currently using supercedent pipeline manager | 17:59 |
| pabelanger | I need to look at scheduler logs to see why C was assigned the node first | 18:00 |
| corvus | pabelanger: are these all for the same project? | 18:01 |
| *** jamesmcarthur has quit IRC | 18:01 | |
| corvus | pabelanger: and are they all for the same branch? | 18:02 |
| pabelanger | corvus: yes | 18:02 |
| pabelanger | and yes | 18:02 |
| corvus | pabelanger: did C get enqueued before A reported? | 18:02 |
| pabelanger | corvus: let me look in logs | 18:03 |
| pabelanger | corvus: yes, C was enqueued before A reported: http://paste.openstack.org/show/748950/ and B was before C | 18:13 |
| corvus | pabelanger: then B should have been dequeued | 18:13 |
| pabelanger | okay, it wasn't, let me see why that is | 18:13 |
| *** jamesmcarthur has joined #zuul | 18:17 | |
| corvus | pabelanger: oh, supercedent matches change and ref | 18:17 |
| corvus | pabelanger: so it won't have determined that C should supercede B | 18:18 |
| pabelanger | Ah, okay | 18:18 |
| corvus | pabelanger: you can see it's behavior by looking for the log messages "Found existing queue" and "Dynamically created queue" | 18:19 |
| corvus | pabelanger: my guess is you will find that each of the 3 changes resulted in "Dynamically created queue" | 18:19 |
| corvus | pabelanger: if so, then it means all 3 changes were running independently | 18:20 |
| corvus | (and the ordering was just nodepool randomness) | 18:20 |
| corvus | pabelanger: i suspect we need to have the supercedent manager check 'branch' if it's available, and if not, ref. | 18:20 |
| corvus | that way it should work as expected in both 'promote' and 'post' pipelines | 18:21 |
| pabelanger | yes, Dynamically created queue is what I see | 18:21 |
| pabelanger | okay, cool. I'll take care for now and rate-limit myself, and try and work on a patch to fix properly | 18:22 |
| pabelanger | corvus: thanks for info! | 18:22 |
| *** jamesmcarthur has quit IRC | 18:25 | |
| corvus | pabelanger: cool -- it's probably a one-liner to supercedent.py; but we'll need a new test case for it | 18:25 |
| pabelanger | yah, I suspect it will take me some time to figure out the testing | 18:26 |
| pabelanger | I try over the weekend | 18:26 |
| pabelanger | I'll* | 18:26 |
| corvus | SpamapS: ^ you may want to be aware of this behavior; we should be able to fix it soon, but in the mean time, you may want to keep an eye on promote | 18:26 |
| *** jamesmcarthur has joined #zuul | 18:29 | |
| *** jamesmcarthur has quit IRC | 18:32 | |
| *** jamesmcarthur has joined #zuul | 18:34 | |
| openstackgerrit | Merged openstack-infra/zuul master: Tolerate missing project https://review.openstack.org/579872 | 18:43 |
| *** bhavikdbavishi has quit IRC | 18:53 | |
| SpamapS | corvus: to be certain I understand, is the negative effect that I run promote jobs more often than I should, because they're not supercedeing right? | 19:05 |
| *** hashar has joined #zuul | 19:16 | |
| corvus | SpamapS: yes, but also, that they could run out of order | 19:17 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: WIP: Support Ansible 2.8 https://review.openstack.org/631933 | 19:19 |
| openstackgerrit | Merged openstack-infra/nodepool master: Remove unused use_taskmanager flag https://review.openstack.org/649293 | 19:58 |
| openstackgerrit | Merged openstack-infra/nodepool master: Remove reference to TaskManagerStopped exception https://review.openstack.org/649294 | 20:02 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: WIP: Support Ansible 2.8 https://review.openstack.org/631933 | 20:03 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: WIP: Remove support for ansible 2.5 https://review.openstack.org/650431 | 20:03 |
| SpamapS | corvus: oh... hm, that's not good | 20:34 |
| SpamapS | I had assumed that it would only ever leap forward in time. | 20:34 |
| corvus | that's the way it's supposed to work. | 20:34 |
| SpamapS | corvus: would independent work the way we want, but just with the down side that it would run all things, not just "the last thing"? | 20:45 |
| SpamapS | Because my promote job is pretty fast... 2 minutes.. now that it doesn't have to build images and upload them. | 20:46 |
| clarkb | independent could still run them out of order depending on node assignments | 20:46 |
| SpamapS | dependent+sempahore would work, yeah? | 20:48 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Check branches in the supercedent pipeline https://review.openstack.org/650440 | 20:49 |
| corvus | SpamapS, pabelanger, clarkb: ^ | 20:50 |
| clarkb | corvus: I think there is a bug in that (left a comment inline) | 20:56 |
| corvus | clarkb: i was relying on the idea that a pipeline probably isn't going to get both kinds of changes, but you are technically correct, which is the best kind, so i'll update :) | 20:58 |
| clarkb | ya the config doesn't prevent triggering a pipeline on both | 20:58 |
| clarkb | whether or not you really want to? I don't know :) | 20:58 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Check branches in supercedent pipelines https://review.openstack.org/650440 | 20:59 |
| corvus | yeah, it doesn't make sense to me but that clearly shouldn't be the limiting factor :) | 20:59 |
| SpamapS | Yeah if we have to have everything in zuul make sense to us... rm -rf zuul/driver/github ;-) | 21:02 |
| *** jamesmcarthur has quit IRC | 21:15 | |
| *** CrayZee has joined #zuul | 21:56 | |
| *** rlandy_ has joined #zuul | 21:56 | |
| *** manjeets_ has joined #zuul | 21:58 | |
| *** tobiash_ has joined #zuul | 21:59 | |
| *** edmondsw has quit IRC | 22:01 | |
| *** rlandy|ruck has quit IRC | 22:01 | |
| *** EmilienM has quit IRC | 22:01 | |
| *** tristanC has quit IRC | 22:01 | |
| *** tobiash has quit IRC | 22:01 | |
| *** sshnaidm|off has quit IRC | 22:01 | |
| *** frickler has quit IRC | 22:01 | |
| *** shachar has quit IRC | 22:01 | |
| *** dkehn has quit IRC | 22:01 | |
| *** manjeets has quit IRC | 22:01 | |
| *** mhu has quit IRC | 22:01 | |
| *** SotK has quit IRC | 22:01 | |
| *** bstinson has quit IRC | 22:01 | |
| *** tristanC_ has joined #zuul | 22:01 | |
| *** frickler has joined #zuul | 22:01 | |
| *** edmondsw_ has joined #zuul | 22:02 | |
| *** EmilienM has joined #zuul | 22:02 | |
| *** SotK has joined #zuul | 22:02 | |
| *** sshnaidm|off has joined #zuul | 22:02 | |
| *** bstinson has joined #zuul | 22:11 | |
| *** rlandy_ has quit IRC | 22:18 | |
| *** openstackgerrit has quit IRC | 22:28 | |
| pabelanger | corvus: thanks! post +2 | 22:54 |
| *** pcaruana has quit IRC | 22:57 | |
| pabelanger | Hmm, unauthorized: incorrect username or password: http://logs.openstack.org/40/650440/2/gate/zuul-upload-image/99cb866/job-output.txt.gz#_2019-04-05_22_31_51_709722 | 22:58 |
| pabelanger | that looks to be coming back from docker.io | 22:59 |
| corvus | no that's from the intermediate or buildset registry | 23:00 |
| corvus | it's happened twice, both on limestone | 23:00 |
| corvus | so i think we still have an ipv6 problem with those roles | 23:00 |
| pabelanger | ack | 23:00 |
| clarkb | possible that docker won't speak ipv6 unless it is enabled globally? | 23:01 |
| clarkb | I know for containers you have to turn it on | 23:01 |
| clarkb | maybe that leaks to the http GET of image data? | 23:01 |
| clarkb | or the registry? | 23:01 |
| logan- | http://logs.openstack.org/40/650440/2/gate/zuul-upload-image/99cb866/job-output.txt#_2019-04-05_22_32_18_606947 | 23:01 |
| clarkb | oh may just need [] escapes | 23:02 |
| corvus | huh, i could have sworn we fixed that | 23:02 |
| corvus | we may have 2 problems there too | 23:03 |
| clarkb | ipwrap is the ansible filter that will auto escape for us | 23:03 |
| clarkb | it will handle ipv4 addrs properly so we should be able to do addr | ipwrap in that jinja template safely | 23:04 |
| corvus | cool, so we'll need that in the push-to-intermediate registry role to fix the problem logan pointed to | 23:04 |
| corvus | and presumably some other place to fix the problem pabelanger pointed to | 23:05 |
| corvus | or perhaps clarkb's theory for pabelanger's problem | 23:05 |
| corvus | it may be the same problem with use-buildset-registry | 23:06 |
| corvus | i'll work on a change | 23:06 |
| corvus | i wonder if certs.d needs similar escaping? | 23:08 |
| *** hashar has quit IRC | 23:10 | |
| *** openstackgerrit has joined #zuul | 23:11 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Use ipwrap filter in registry roles https://review.openstack.org/650473 | 23:11 |
| corvus | clarkb, logan-, pabelanger: ^ that's all the places where urls are used, i also included the user config file for docker (where the passwords are stored). it's unclear whether we shoud be escaping there. I also don't know if we need to escape the directory names in /etc/certs.d. | 23:11 |
| clarkb | I think you'll need ti anywhere the port is included | 23:12 |
| clarkb | because its the :5000 suffix that confuses stuff | 23:12 |
| corvus | that is going to be a *very* messy directory name | 23:12 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Use ipwrap filter in registry roles https://review.openstack.org/650473 | 23:15 |
| corvus | okay that's every occurance of "registry.host" | 23:15 |
| corvus | clarkb: will ipwrap handle hostnames okay? | 23:15 |
| corvus | (ie, will it ignore them?) | 23:15 |
| clarkb | corvus: https://docs.ansible.com/ansible/latest/user_guide/playbooks_filters_ipaddr.html#wrapping-ipv6-addresses-in-brackets are the docs. It says it will ignore them | 23:16 |
| clarkb | "It will wrap all IPv6 addresses and leave any other strings intact:" | 23:16 |
| corvus | okay cool, then it should be safe to apply to intermediate_registry.host (which that patch does) | 23:16 |
| corvus | that should be a hostname for us | 23:16 |
| corvus | (but it'd be nice not to require that, so i think using it is a good choice) | 23:17 |
| pabelanger | should we depends-on test that or is there a self test? | 23:21 |
| clarkb | pabelanger: I think that jobs that use it are in trusted repo so has to be merged first | 23:22 |
| pabelanger | kk, I left +2, due to weekend, but if somebody else wants to +3 | 23:22 |
| corvus | yeah, sorry i can't shepherd that one through; unfortunately without that fixed, zuul and nodepool are going to hit sporadic failures due to it, but because of the trusted repo thing, we need to merge it and verify it's working.... i'm about to be afk for a week, so it'd be great if someone could pick that up next week. | 23:29 |
| corvus | i think i know enough about this now to make test jobs for those roles... i'll try to do that after i get back. that still doesn't help us now. :/ | 23:30 |
| *** needscoffee is now known as kmalloc | 23:32 | |
| SpamapS | Hm I'm going to need to see if I can make AWS's ECR work like the buildset registry. | 23:52 |
| SpamapS | More and more jobs need access to a docker image that should only need to build once, even in check | 23:53 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!