| *** panda has quit IRC | 00:13 | |
| *** panda has joined #zuul | 00:16 | |
| tristanC | Shrews: do you think the test_kubernetes.py should also start the launcher? or should we adapt setup_config() to kill and restart the service? | 00:50 |
|---|---|---|
| tristanC | tobiash: do you have console logs about serviceworker causing issue? is the ui served over tls? | 00:53 |
| tristanC | corvus: about the "quick-start test job", shouldn't it also use tox for validation? or could it be designed so that it's easy to use for other purpose, e.g. a role that "setup-job zuul_yaml=... playbook_file=..." ? | 00:55 |
| tristanC | corvus: are the html files in the container? | 00:56 |
| tristanC | mordred: it would be easier if the quickstart was driven by ansible instead of docker-compose... | 00:57 |
| *** clarkb has quit IRC | 01:38 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add version to info endpoint https://review.openstack.org/609571 | 01:41 |
| *** tflink has quit IRC | 01:47 | |
| *** tflink has joined #zuul | 01:50 | |
| Shrews | tristanC: the useNodepool() call starts the launcher | 02:11 |
| Shrews | that's why i removed the launcher start from the playbook | 02:12 |
| Shrews | (at least, i think i did. it's late) | 02:13 |
| tristanC | Shrews: i don't see the removal... | 02:13 |
| tristanC | Shrews: anyway, do you prefer if the launcher service is managed by tox unittest instead of long-running service? | 02:13 |
| Shrews | tristanC: starting it in tox is the only way to start with different nodepool configs, which is necessary for different test scenarios | 02:14 |
| tristanC | how should we pass environment variable like kube context name? | 02:14 |
| dmsimard | tristanC: hum, vars that would be in a dict which we'd run through and run export from ? or set the environment for every task | 02:15 |
| Shrews | i don't know what that env var is, but tox has a way to pass in env vars (passenv) | 02:15 |
| dmsimard | oh wait, I didn't read far back enough in the backlog :p | 02:15 |
| tristanC | Shrews: i meant from zuul-job playbook to the tox process env, perhaps the tox role is missing an environment variable? | 02:16 |
| tristanC | Shrews: context name is similar to OS_CLOUD variable | 02:16 |
| Shrews | tristanC: i'm not clear what you mean, but what i put up is by no means a final solution. just something to get us started | 02:18 |
| Shrews | a POC | 02:19 |
| * Shrews needs to head off to bed now | 02:20 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: builds: set missing start/end time for node errors https://review.openstack.org/535549 | 02:23 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix stream url from status json https://review.openstack.org/609365 | 02:34 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: remove legacy stream.html url support https://review.openstack.org/609366 | 02:34 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: use window.location in TableFilters to keep sub-url path https://review.openstack.org/609578 | 02:47 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add job page https://review.openstack.org/597048 | 02:53 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add config-errors notifications drawer https://review.openstack.org/597147 | 02:53 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add change status page https://review.openstack.org/599472 | 02:55 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add build page https://review.openstack.org/597024 | 02:57 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add project page https://review.openstack.org/604266 | 02:58 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add job page https://review.openstack.org/597048 | 02:58 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add config-errors notifications drawer https://review.openstack.org/597147 | 03:15 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add change status page https://review.openstack.org/599472 | 03:15 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add project page https://review.openstack.org/604266 | 03:15 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add labels page https://review.openstack.org/604682 | 03:15 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add nodes page https://review.openstack.org/604683 | 03:15 |
| *** sshnaidm|afk has quit IRC | 05:19 | |
| *** chkumar|off is now known as chandankumar | 05:26 | |
| tobiash | tristanC: according to the network tab in the browser it was serviced by the serviceworker | 05:28 |
| tobiash | tristanC: and that was the problem because our authentication relies on redirects to the auth provider which doesn't happen if the main page comes from the service worker -> api requests broken | 05:28 |
| tobiash | tristanC: so now that I know that I need to disable the service worker I need a clean way to disable it (probably at build time). Do you have any idea other than patching a 'return' into it? | 05:30 |
| tristanC | tobiash: you can remove it from the index.js | 05:34 |
| tristanC | i don't understand the issue, the serviceworker shouldn't interfer with redirections | 05:35 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix stream url from status json https://review.openstack.org/609365 | 05:37 |
| tobiash | tristanC: the issue is that I need the redirection for authentication (which is done when fetching the index.html from the server), the serviceworker skips that so the redirect happens on the first request to the server | 05:38 |
| tobiash | which happens to be an api request which then gets redirected to a login page | 05:39 |
| tristanC | tobiash: perhaps we could handle that in this first api request handler, e.g. if it's a redirect, then change the window location accordingly | 05:40 |
| tobiash | tristanC: so redirect the whole page to the login page? | 05:40 |
| tobiash | tristanC: if we can make this working it probably would be the best solution | 05:41 |
| tristanC | tobiash: it may be possible to that here: https://git.zuul-ci.org/cgit/zuul/tree/web/src/reducers.js#n57 | 05:41 |
| tristanC | by inspecting the response content | 05:41 |
| *** quiquell|off is now known as quiquell | 05:42 | |
| tristanC | or even dispatch a LOGIN_REDIRECT_SUCCESS action that would trigger a redirect, or even load the login page in an overlay | 05:43 |
| tobiash | tristanC: so maybe we can tell it to not follow redirects and detect this situation? | 05:44 |
| *** swest has joined #zuul | 05:47 | |
| tobiash | tristanC: we could set maxRedirects to 0 when using Axios.get: https://github.com/axios/axios/blame/master/README.md#L343 | 05:47 |
| tristanC | yes | 05:47 |
| tobiash | that should hopefully return the 30x code which we can intercept and manually do the redirect | 05:47 |
| tobiash | (or refresh the page with invalidated cache?) | 05:48 |
| tobiash | ok, but first step is to try this and log the redirect... | 05:48 |
| * tobiash rereads docs to find out how to run/debug that locally | 05:50 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Fix broken code block in the docs https://review.openstack.org/609590 | 05:55 |
| *** pcaruana has joined #zuul | 06:01 | |
| tobiash | hrm, maxRedirects doesn't work in the browser :/ | 06:49 |
| *** goern has joined #zuul | 06:53 | |
| *** frickler has joined #zuul | 07:16 | |
| *** ssbarnea has joined #zuul | 07:17 | |
| frickler | it seems that zuul has trouble handling changes to job definitions in the case of merge requests https://review.openstack.org/609455 | 07:17 |
| frickler | the merge removes two jobs from .zuul.yaml and their playbooks, but zuul still seems to try to run these jobs | 07:18 |
| *** slaweq has joined #zuul | 07:22 | |
| frickler | corvus: tobiash: does that ring a bell for you maybe? ^^ | 07:24 |
| tobiash | frickler: that is the case because gerrit doesn't tell zuul that the zuul config has been changed | 07:27 |
| tobiash | frickler: we would need a similar solution like https://review.openstack.org/603287 to resolve this | 07:28 |
| tobiash | frickler: zuul gets the changed files from the change metadata and gerrit doesn't list zuul.yaml there | 07:29 |
| frickler | tobiash: ah, that makes sense. is this a known issue or would you like me to create a story for it? | 07:31 |
| tobiash | I don't think there is a story for that already | 07:31 |
| frickler | slaweq: maybe you want to do this yourself? ^^ | 07:33 |
| *** ssbarn___ has joined #zuul | 07:33 | |
| slaweq | frickler: tobiash: but want do what exactly? | 07:35 |
| *** themroc has joined #zuul | 07:36 | |
| frickler | slaweq: create a story for our issue at https://storyboard.openstack.org/#!/project/679 | 07:36 |
| slaweq | frickler: ahh, sure | 07:37 |
| slaweq | I will create it in few minutes | 07:37 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: remove legacy stream.html url support https://review.openstack.org/609366 | 07:59 |
| *** ssbarnea has quit IRC | 08:13 | |
| slaweq | frickler: tobiash: I created story for this issue https://storyboard.openstack.org/#!/story/2004046 | 08:14 |
| *** AJaeger has quit IRC | 08:15 | |
| slaweq | tobiash: do You know if there is any other way how we can make this merge to be done without waiting for proper fix in zuul? | 08:16 |
| *** AJaeger has joined #zuul | 08:25 | |
| tobiash | slaweq: one way would be to land the changes instead of the merge commit | 08:30 |
| tobiash | (if possible) | 08:30 |
| slaweq | tobiash: so doing cherry-picks of each commit to this feature branch, right? | 08:31 |
| tobiash | slaweq: yes, that's the idea (but not a good one if that consists of many commits) | 08:31 |
| slaweq | yes, it's few months to be synced - even if I would click all of that, it would kill the gate for some time :) | 08:32 |
| slaweq | that is not an option here :/ | 08:32 |
| tobiash | slaweq: if the problem is just removed jobs, you could land just the zuul.yaml change and after that the merge | 08:33 |
| slaweq | tobiash: that may be an option here | 08:33 |
| slaweq | tobiash: I will try :) | 08:33 |
| slaweq | tobiash: thx | 08:34 |
| tobiash | :) | 08:34 |
| tobiash | slaweq: or you amend the merge commit and add a newline to zuul.yaml | 08:34 |
| tobiash | that might also work | 08:34 |
| tobiash | the important thing is just that gerrit lists zuul.yaml as changed | 08:34 |
| slaweq | I think that cherry-picks of 2 commits would be "cleaner" solution | 08:35 |
| tobiash | slaweq: as you like, both ways should work | 08:36 |
| tobiash | tristanC: it looks like the least hacky option is to disable the service worker | 08:38 |
| tobiash | tristanC: and I just learned how to do that with a compile time switch :) | 08:38 |
| frickler | slaweq: another option would be to force-merge the request in gerrit, circumventing zuul | 08:42 |
| *** electrofelix has joined #zuul | 08:42 | |
| slaweq | frickler: who can do that? PTL has such rights maybe? | 08:42 |
| *** ssbarnea has joined #zuul | 08:49 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Optionally disable service worker in zuul-web https://review.openstack.org/609626 | 08:51 |
| tobiash | tristanC, mordred, corvus: this fixes the zuul-web fallout we had ^ | 08:51 |
| frickler | slaweq: any infra-root can do it, so I'd volunteer, but it would probably be good to get PTL confirmation for that indeed | 08:51 |
| *** ssbarnea has quit IRC | 08:56 | |
| tristanC | tobiash: with the tenant scoped admin api, we should be able to accomodate login requirements from the ui | 08:59 |
| *** panda is now known as panda|mtg | 09:05 | |
| *** ssbarn___ has quit IRC | 09:08 | |
| *** ssbarnea has joined #zuul | 09:19 | |
| *** AJaeger_ has joined #zuul | 09:37 | |
| *** AJaeger has quit IRC | 09:40 | |
| *** AJaeger_ is now known as AJaeger | 09:41 | |
| *** jiapei has left #zuul | 09:46 | |
| *** gouthamr has quit IRC | 10:14 | |
| *** dmellado has quit IRC | 10:17 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Make update intercal configurable at compile time https://review.openstack.org/609664 | 10:44 |
| *** sshnaidm|afk has joined #zuul | 10:45 | |
| tobiash | tristanC: yes, that might be a mid to longterm solution but this also needs thinking how this can be integrated to various auth frameworks then | 10:51 |
| *** sshnaidm|afk is now known as sshnaidm | 11:02 | |
| slaweq | frickler: thx, so I will ask mlavalle to get back to You or anyone else from infra-root | 11:03 |
| *** quiquell is now known as quiquell|afk | 11:15 | |
| tobiash | tristanC: it looks like you don't have to patch zuul to host it under /zuul/ | 11:16 |
| tobiash | tristanC: building with 'PUBLIC_URL='/zuul/' yarn build' seems to produce the same correct result | 11:16 |
| tobiash | tristanC: was there a reason to delete the jshint file? It helped my IDE to select the correct warnings profile before. | 11:20 |
| tristanC | tobiash: that's good to know, even simpler to document then | 11:26 |
| tristanC | tobiash: no reason, we can restore it | 11:26 |
| tobiash | tristanC: was just about to document this (after testing that it really works) | 11:26 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: use PUBLIC_URL instead of package.json patch https://review.openstack.org/609672 | 11:28 |
| *** jpena|off has joined #zuul | 12:02 | |
| *** gouthamr has joined #zuul | 12:05 | |
| *** dmellado has joined #zuul | 12:10 | |
| *** quiquell|afk is now known as quiquell|lunch | 12:11 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: executor: enable zuul_return to update Ansible inventory https://review.openstack.org/590092 | 12:17 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: executor: add support for generic build resource https://review.openstack.org/570668 | 12:17 |
| mordred | tobiash: 609664 lgtm - but I also wonder - would it be useful for us to make that a zuul.conf setting exposable via the info endpoint? | 12:20 |
| *** rlandy has joined #zuul | 12:32 | |
| tristanC | mordred: agreed, now that we load the info endpoint by default each time the dashboard load, it could be used for such settings in context store | 12:37 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix rewrite rule installation doc https://review.openstack.org/608850 | 12:39 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Make update interval configurable at compile time https://review.openstack.org/609664 | 12:41 |
| tobiash | mordred: I'm fine with either, this way was just the easiest for me after I discovered how to fix the service worker ;) | 12:42 |
| mordred | tobiash: ++ | 12:42 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: [WIP] web: enable status update interval set in info endpoint. https://review.openstack.org/609690 | 12:46 |
| tristanC | tobiash: that's how it may work from /info endpoint ^ | 12:46 |
| *** quiquell|lunch is now known as quiquell | 12:47 | |
| tobiash | tristanC: cool, that's probably the better option | 12:54 |
| tobiash | tristanC: btw, you never sleep, do you? | 12:55 |
| tobiash | ;) | 12:55 |
| tristanC | i do sleep very well, it's just my timezone that is hard to work with ;) | 13:06 |
| mordred | tristanC: I agree with tobiash - I choose to believe you do not sleep | 13:09 |
| mhu | tristanC is not bound by time or space | 13:21 |
| *** quiquell is now known as quiquell|off | 13:59 | |
| *** ssbarne__ has joined #zuul | 14:03 | |
| *** panda|mtg has quit IRC | 14:12 | |
| *** panda has joined #zuul | 14:14 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: web: use PUBLIC_URL instead of package.json patch https://review.openstack.org/609672 | 14:33 |
| corvus | tobiash: i'm curious about 609664 -- i set it to 5s because i didn't think it would be worth updating any faster than that at any scale (and status updates have an impact on scheduler loads). considering that jobs take minutes to run, does it make that much difference? and what caused you to decide it was too slow? | 14:46 |
| tobiash | corvus: on a smaller zuul it just felt snappier when the items show up 2s after the pull request. Actually it was a user request who had the 'feeling' that zuul is slow and to pretend a faster zuul I patched this months ago | 14:48 |
| tobiash | today I decided to either upstream or drop this feature... | 14:48 |
| corvus | tobiash: you set it to 2s? | 14:49 |
| tobiash | yes | 14:49 |
| tobiash | corvus: and I just abandoned the change and will go back to 5s for now | 14:49 |
| tobiash | let's see what my users say (if anything) | 14:49 |
| corvus | in the react dashboard, how is the interval measured? from the start of each request, or from the end? | 14:49 |
| corvus | tobiash: ok -- to be clear, it's only a -1 review, not a -2 -- we can change my mind. :) but i do want to avoid adding configuration options which aren't necessary (and i especially want to avoid configuration options which aren't optimal by default) | 14:51 |
| tobiash | it looks to me like a pause from the end | 14:51 |
| tobiash | corvus: well I think it's best to not invest time on this until I may get complaints from users | 14:51 |
| corvus | tobiash: from the end probably makes it a bit safer to decrease the default -- it means that the absolute interval will scale with response time. | 14:51 |
| corvus | tobiash: okay. we can file it away for later then | 14:52 |
| tobiash | *if* I get complaints from users I will re-open the discussion on this :) | 14:52 |
| corvus | also, what i *really* want to do is use a websocket to emit events that send only update events in real-time. but that's probably zuul v4 (queue information in zookeeper). | 14:53 |
| tobiash | corvus: also the implementation would be better with the info endpoint as a real config option instead of a compile time option so the way of implementing this was the wrong way anyway | 14:53 |
| corvus | tobiash: yeah, i saw the other change and agree that's better if we make it configurable. | 14:53 |
| fungi | for those interested, there's some discussion with coreycb ni #openstack-infra about stubbing out an initial ensure-python role in zuul-jobs we can add to things like the tox pre playbook so that tox jobs can make sure an appropriate python version is present (needed for py37 testing on ubuntu bionic) | 14:54 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix broken code block in the docs https://review.openstack.org/609590 | 14:54 |
| tobiash | regarding the other change I think this needs to be compile time because that's very early and also probably a very specific need from us | 14:54 |
| corvus | tobiash: yeah -- though i thought tristanC said you might be able to detect it in js and handle it? i think if that works, it would be better -- i actually expect a lot of auth systems would have this problem. but i don't see a problem with landing the compilation flag until we figure that out. | 14:57 |
| tobiash | corvus: I tried that but I found out that there is no way to detect this reliably in js so I went ahead and implemented the second best option (which is also the simplest) | 14:58 |
| *** openstackgerrit has quit IRC | 14:58 | |
| tobiash | corvus: my hope was to limit the redirects to 0 but that only works in nodejs, browsers don't give a choice on that | 14:58 |
| tobiash | corvus: the 'real' solution is probably built-in auth support in zuul-web, but that's a longer story to discuss | 14:59 |
| corvus | tobiash: what happens when the browser gets the redirect? do we get the login page returned as if it were the api end point? | 14:59 |
| tobiash | corvus: when the browser gets the redirect it depends on the type of request, a document request just follows the redirect and presents the login screen | 15:00 |
| corvus | tobiash: mhu is ready to discuss it, and i think we'll be ready to join the discussion once we get the current dashboard and container stuff out of the way :) | 15:00 |
| *** openstackgerrit has joined #zuul | 15:00 | |
| tobiash | corvus: an api request is followed and will return a login screen (or xss failure) to the component that expects json | 15:01 |
| corvus | tobiash: so, theoretically, we should be able to detect that and do something with it? | 15:01 |
| tobiash | so theoretically we can detect broken json or a generic exception | 15:02 |
| tobiash | but that first nothing we could follow manually and second would hide real errors which I consider a hack | 15:02 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 15:05 |
| *** ssbarne__ has quit IRC | 15:07 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Fixes for quick-start playbooks https://review.openstack.org/609551 | 15:08 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Remove disable-recommends in quick-start test https://review.openstack.org/609553 | 15:08 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Quick-start: remove ansible Dockerfile https://review.openstack.org/609555 | 15:08 |
| *** clarkb has joined #zuul | 15:10 | |
| *** openstackgerrit has quit IRC | 15:22 | |
| *** themroc has quit IRC | 15:23 | |
| *** mugsie has joined #zuul | 15:24 | |
| *** snk has joined #zuul | 15:25 | |
| corvus | actually... it's looking a lot like the delay is from the *start* of the request | 15:29 |
| corvus | because right now, i'm looking at the network panel for the openstack dashboard, and it's starting a new request every 5 seconds, even though it's taking 40 seconds to get a response | 15:29 |
| corvus | (which may have something to do with why it's taking 40s to get a response) | 15:30 |
| *** chandankumar is now known as chkumar|off | 15:32 | |
| *** openstackgerrit has joined #zuul | 15:34 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Add tox functional testing for drivers https://review.openstack.org/609515 | 15:34 |
| corvus | tristanC: do you think we could go back to only having one outstanding status request? | 15:35 |
| mordred | corvus: ++ | 15:46 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Add tox functional testing for drivers https://review.openstack.org/609515 | 16:09 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Do only one status request at a time https://review.openstack.org/609757 | 16:10 |
| tobiash | corvus: that fixes the multiple requests at the same time ^ | 16:11 |
| tobiash | (tested locally against openstack with 500ms interval) | 16:11 |
| corvus | that looks plausible :) | 16:12 |
| tobiash | tristanC, mordred ^ | 16:12 |
| tobiash | corvus: do we want to continue polling on a failed request or stop the loop? | 16:12 |
| tobiash | I think current behavior is continue polling, if we want to keep that I need to push up a quick update | 16:13 |
| corvus | tobiash: yeah, i think we should continue. we want to keep working across a zuul-web restart or transient error | 16:13 |
| mordred | agree | 16:14 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Do only one status request at a time https://review.openstack.org/609757 | 16:14 |
| tobiash | that should do it ^ | 16:14 |
| mordred | looks great | 16:16 |
| tobiash | yay I did zuul-web patches... and it really feels easier than before :) | 16:17 |
| clarkb | the performance is particularly noticeable on mobile | 16:18 |
| *** caphrim007 has joined #zuul | 16:20 | |
| *** panda has quit IRC | 16:20 | |
| corvus | mordred, tristanC: /usr/local/lib/python3.7/site-packages/zuul/web/static is still empty on the zuul-web container images | 16:20 |
| corvus | or nearly empty: http://paste.openstack.org/show/731906/ | 16:21 |
| *** panda has joined #zuul | 16:21 | |
| mordred | tobiash: yay! | 16:21 |
| mordred | corvus: thanks for reminding me - I knew there was a pressing thing I should look in to | 16:22 |
| corvus | mordred: i think we thought we fixed it but i guess not | 16:22 |
| mordred | corvus: well - yeah. I mean - I feel like I definitely fixed it once | 16:22 |
| mordred | but maybe I didn't hit it hard enough and the lid came bac off | 16:23 |
| corvus | mordred: before you dig into that -- i have a qq | 16:23 |
| mordred | corvus: kk | 16:23 |
| corvus | mordred: i'm adding a job which runs docker-compose (https://review.openstack.org/609514) do you have any hints on how we could build and use new images in that job (rather than using already published ones)? | 16:24 |
| mordred | corvus: yes! | 16:24 |
| mordred | corvus: just run pbrx in the zuul and nodepool git repos - which will build images of the corectname in the local docker image cache | 16:25 |
| mordred | corvus: then docker-compose should use the local images since they will exist | 16:25 |
| corvus | mordred: oh neat. and i can test that locally too. excellent, thanks! | 16:25 |
| mordred | corvus: you'll want to give it the zuul prefix so that it tags them properly | 16:26 |
| corvus | ack | 16:26 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Quick-start: remove ansible Dockerfile https://review.openstack.org/609555 | 16:26 |
| *** gouthamr has quit IRC | 16:34 | |
| *** dmellado has quit IRC | 16:34 | |
| openstackgerrit | Corey Bryant proposed openstack-infra/zuul-jobs master: Add ensure-python role https://review.openstack.org/609761 | 16:41 |
| *** gouthamr has joined #zuul | 16:43 | |
| corvus | mordred: what do you think about this for a next step: pause pbrx_build_container_images and have the quick-start job run after it and use its image service to get pre-built images | 16:44 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Build new container images for quick-start https://review.openstack.org/609763 | 16:45 |
| mordred | corvus: I think I'm too jetlagged to fully understand that sentence | 16:45 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Use zuul/web/static in setup_hook https://review.openstack.org/609764 | 16:48 |
| mordred | corvus: fwiw, I'm firing shots in the dark here | 16:48 |
| mordred | python setup.py bdist_wheel is totally doing the right thing for me locally | 16:49 |
| openstackgerrit | Chandan Kumar proposed openstack-infra/zuul master: Added steps to install docker for all distros https://review.openstack.org/609765 | 16:49 |
| chkumar|off | corvus: mordred ^^ | 16:49 |
| mordred | chkumar|off: for the first ones - especially debuntu - docker-compose should be depending on docker | 16:50 |
| mordred | did you experience that not being true? | 16:51 |
| chkumar|off | mordred: on fedora docker-compose does not install docker | 16:51 |
| tobiash | corvus: pausing and getting the images from a different node is probably a good idea once we have provider stickyness in nodepool | 16:52 |
| mordred | chkumar|off: awesome | 16:52 |
| clarkb | mordred: docker-compose only recommends docker.io on debuntu so for most peopel it is installed by default (but not on our test machiens) | 16:52 |
| clarkb | also the package is docker.io on debuntu | 16:52 |
| openstackgerrit | Chandan Kumar proposed openstack-infra/zuul master: Added steps to install docker for all distros https://review.openstack.org/609765 | 16:52 |
| mordred | it's docker-ce | 16:52 |
| clarkb | mordred: only if you add the docker repo from upstream | 16:53 |
| tobiash | corvus: until we have that I think it might be better to build the images in the same job as using them | 16:53 |
| mordred | from the upstream repo | 16:53 |
| mordred | yeah | 16:53 |
| clarkb | mordred: those directions don't use the upstream repo | 16:53 |
| clarkb | so package is docker.io | 16:53 |
| mordred | clarkb: it's only docker.io on bionic - I believe it's docker on xenial - but this shouldn't be an issue because docker-compose should depend on the right thing on debian/ubuntu | 16:54 |
| mordred | the docker package naming is a total mess in debian land | 16:54 |
| clarkb | mordred: I just checked on xenial it is docker.io there too. the reason for that is docker is some 20 year old package doing something else | 16:54 |
| clarkb | but yes I agree installing docker-compose should do the right thing most of the time | 16:54 |
| mordred | ++ - I think ;et's stick with not putting docker or docker.io in the list for debuntu | 16:55 |
| openstackgerrit | Corey Bryant proposed openstack-infra/zuul-jobs master: Add ensure-python role https://review.openstack.org/609761 | 16:56 |
| corvus | tobiash: i'm not too worried about transferring images between providers at this point. they aren't huge. and maybe the lower layers can come from the local dockerhub cache? i dunno if it works that way but it should. :) | 16:56 |
| mordred | corvus: ah - gotit. well- at the very least I think it's worth trying, even if it's unhappy we'll likely learn somthing about what seems to be a common desire | 16:58 |
| tobiash | corvus: that's right | 16:58 |
| tobiash | corvus: do you know how you want to serve the images on the node that builds them? | 16:58 |
| mordred | chkumar|off: sorry - I left comments on PS#1 - you beat me to PS#2 before I clicked the button | 16:58 |
| corvus | about the service? i was just about to ask about that | 16:58 |
| openstackgerrit | Merged openstack-infra/zuul master: web: use window.location in TableFilters to keep sub-url path https://review.openstack.org/609578 | 16:59 |
| tobiash | corvus: the easiest thing would be to start a local registry using this: https://hub.docker.com/_/registry/ | 16:59 |
| corvus | tobiash: yeah, so we might need to add that as an option to the pbrx build job | 17:00 |
| tobiash | corvus: but you need to take care of the prefix sp you can push them into it | 17:00 |
| tobiash | maybe a hosts entry simplifies that | 17:00 |
| clarkb | and set up auth of some sort | 17:00 |
| chkumar|off | mordred: upating the patchset | 17:00 |
| clarkb | so that someone else can't pollute your job | 17:00 |
| mordred | clarkb: they'd have a moving target to try to push things to ... you think that's likely enough to require setting up auth? | 17:01 |
| corvus | okay so not a 5 minute task :) but i think we've got all the parts to start building that | 17:01 |
| tobiash | correct summary ;) | 17:01 |
| clarkb | mordred: to start maybe not? but I'd be wary of relying on test results that can be attacekd in that manner | 17:02 |
| corvus | yeah.... i think as soon as this works for tests, it won't be long till we have promotion/publication pipelines based on this, and at that point it's probably important. | 17:02 |
| tobiash | corvus: we have roles of a working (unauthenticated) buildset registry in our private repos | 17:03 |
| tobiash | corvus: I can check if I can upstream parts or all of this to zuul-jobs | 17:03 |
| corvus | tobiash: that'd be great, thanks | 17:03 |
| corvus | i'll restrain myself from working on this for a few days :) | 17:04 |
| *** electrofelix has quit IRC | 17:13 | |
| *** AJaeger has quit IRC | 17:15 | |
| corvus | i think we can switch the quick-start to use polygerrit instead of the gwt ui | 17:19 |
| corvus | i didn't do it to start with because it's just a little bit more complicated to set up via automation, but i think i've got solutions to those minor problems | 17:20 |
| clarkb | corvus: its a per user setting, not sure if you can set it server wide | 17:20 |
| corvus | that should hopefully make people's eyes bleed less | 17:20 |
| corvus | clarkb: it's a global setting too | 17:20 |
| clarkb | oh nice | 17:20 |
| corvus | clarkb: and actually, whether it's permitted as a user setting is also a global setting | 17:20 |
| clarkb | huh | 17:20 |
| *** AJaeger has joined #zuul | 17:21 | |
| corvus | so i'll go through it again today and update the directions/screenshots for that. we'll be more future-proof too | 17:21 |
| corvus | since i think the next version will be polygerrit by default | 17:21 |
| openstackgerrit | Chandan Kumar proposed openstack-infra/zuul master: Added steps to install dependencies using bindep https://review.openstack.org/609765 | 17:22 |
| chkumar|off | mordred: corvus ^^ | 17:22 |
| mordred | chkumar|off: ++ | 17:22 |
| corvus | i'm sorry, i guess i wasn't clear. i do not think we should install bindep. | 17:23 |
| corvus | that's what y tried to communicate in my comment on the change | 17:23 |
| chkumar|off | corvus: you mean just add install blan blah and start the service? | 17:24 |
| chkumar|off | no use of bindeps | 17:25 |
| corvus | chkumar|off: yep. patchset 2 was almost perfect. it just needed to scope the service commands like monty suggested. | 17:25 |
| chkumar|off | *bindep | 17:25 |
| corvus | chkumar|off: correct. that's way too complicated for this. | 17:25 |
| clarkb | fwiw apt-get install docker-compose did just work for all of that stuff on ubuntu for me | 17:25 |
| clarkb | so I agree having bindep as a dep is probably overkill | 17:25 |
| chkumar|off | clarkb: upadting | 17:25 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Build new container images for quick-start https://review.openstack.org/609763 | 17:30 |
| mordred | oh - sorry - I was trying tp ++ corvus above ... bad inflight wifi | 17:32 |
| corvus | cool, i'm going to afk for a bit, then i'm on the polygerrit stuff | 17:32 |
| openstackgerrit | Chandan Kumar proposed openstack-infra/zuul master: Added steps to install docker for all distros https://review.openstack.org/609765 | 17:33 |
| chkumar|off | mordred: clarkb ^6 | 17:34 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul-jobs master: WIP: Add roles to work with a buildset docker registry https://review.openstack.org/609773 | 17:35 |
| tobiash | corvus: ^ | 17:35 |
| corvus | tobiash: wow that was fast | 17:36 |
| tobiash | just needed some minor tweaks and a good commit message ;) | 17:37 |
| tobiash | we also have a real test job for this but that uses other not yet upstreamed roles and thus needs to be reworked | 17:38 |
| *** ianychoi has quit IRC | 17:41 | |
| chkumar|off | corvus: clarkb mordred for fedora, we need sudo docker-compose | 17:44 |
| clarkb | chkumar|off: yes there is a change for that pushed already | 17:44 |
| chkumar|off | clarkb: thanks :-) | 17:44 |
| tobiash | corvus, clarkb, mordred: what do you think about splitting zuul.yaml in zuul-jobs into several files? | 17:45 |
| tobiash | I think that could make it easier to maintain and extend | 17:45 |
| chkumar|off | clarkb: i think we can one line shell script to do all the stuff | 17:45 |
| chkumar|off | installing git then dependencies and followed by docker-compose | 17:46 |
| chkumar|off | what you say? | 17:46 |
| tobiash | for the docker buildset registry I also would add a test job that does an end-to end test of these roles | 17:46 |
| clarkb | chkumar|off: corvus is the person you want input from but I think this is supposed to be slightly instructional? I don't know. | 17:46 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul-jobs master: WIP: Add roles to work with a buildset docker registry https://review.openstack.org/609773 | 17:58 |
| *** openstackgerrit has quit IRC | 18:20 | |
| *** gouthamr_ has joined #zuul | 18:25 | |
| *** openstackgerrit has joined #zuul | 18:27 | |
| openstackgerrit | Merged openstack-infra/zuul master: web: use PUBLIC_URL instead of package.json patch https://review.openstack.org/609672 | 18:27 |
| *** dmellado has joined #zuul | 18:27 | |
| *** openstackstatus has quit IRC | 18:28 | |
| *** openstackstatus has joined #zuul | 18:31 | |
| *** ChanServ sets mode: +v openstackstatus | 18:31 | |
| openstackgerrit | Corey Bryant proposed openstack-infra/zuul-jobs master: Add ensure-python role https://review.openstack.org/609761 | 18:33 |
| *** panda has quit IRC | 18:35 | |
| *** panda has joined #zuul | 18:38 | |
| *** gouthamr has quit IRC | 18:45 | |
| *** AJaeger_ has joined #zuul | 18:46 | |
| *** AJaeger has quit IRC | 18:48 | |
| *** AJaeger_ is now known as AJaeger | 18:50 | |
| openstackgerrit | Merged openstack-infra/zuul-sphinx master: Use OrderedDict for object tracking https://review.openstack.org/605240 | 19:22 |
| openstackgerrit | Merged openstack-infra/zuul-sphinx master: Add example and type options to attributes https://review.openstack.org/604267 | 19:22 |
| pabelanger | mordred: know you are traveling, but would love to loop back to https://review.openstack.org/583346/ basically the zuulv3-output stuff you started a while back. This was a different approach to your original idea: https://review.openstack.org/511843/ | 19:24 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Add zuul to the linters tox env https://review.openstack.org/607026 | 19:27 |
| openstackgerrit | Merged openstack-infra/zuul master: web: fix multiple builds query parameters for a single column https://review.openstack.org/593605 | 19:28 |
| mrhillsman | is it possible to modify the main.yaml without restarting scheduler? | 19:40 |
| clarkb | mrhillsman: yes there is a reload command that will reload the config without restarting the process | 19:43 |
| tobiash | mrhillsman: yes, just do a full reload: https://zuul-ci.org/docs/zuul/admin/components.html#operation | 19:43 |
| mrhillsman | thx | 19:44 |
| clarkb | looking at puppet-zuul I think we may still use the sighup method | 19:45 |
| tobiash | clarkb: which is deprecated now ;) | 19:46 |
| clarkb | tobiash: yup | 19:46 |
| corvus | clarkb: aha it was host key verification (again): http://logs.openstack.org/14/609514/7/check/zuul-quick-start/2fe2040/ara-report/result/059db004-41cf-4927-9d7a-a00e964541f5/ | 19:48 |
| clarkb | bah | 19:49 |
| clarkb | that particular bug is probably worth documenting in the use a gerrit setup steps | 19:49 |
| corvus | clarkb: i think if there's a terminal, it will ask, so it should be okay. | 19:50 |
| pabelanger | Shrews: we found a bug in the static nodepool driver, working to collect logs, but basically if we lose network access to a static node, nodepool will deregister the node. But, once network is restored, nodepool won't register the static node | 19:52 |
| pabelanger | only way to fix nodepool-launcher is to restart or update configuration file | 19:52 |
| pabelanger | any ideas how to better deal with that? | 19:52 |
| pabelanger | tobiash: I think you might be using static nodes also?^ | 19:53 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 19:53 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Fixes for quick-start playbooks https://review.openstack.org/609551 | 19:53 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Remove disable-recommends in quick-start test https://review.openstack.org/609553 | 19:53 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Quick-start: remove ansible Dockerfile https://review.openstack.org/609555 | 19:53 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Build new container images for quick-start https://review.openstack.org/609763 | 19:53 |
| openstackgerrit | Jeremy Stanley proposed openstack-infra/zuul-base-jobs master: Correct zuul-jobs path https://review.openstack.org/599607 | 19:54 |
| tobiash | pabelanger: I thought we had patches that fix this | 19:57 |
| tobiash | pabelanger: do you have https://review.openstack.org/#/c/600084 in your deployment? | 19:59 |
| pabelanger | tobiash: oh, maybe. | 19:59 |
| pabelanger | tobiash: I am not sure, checking release notes | 19:59 |
| pabelanger | tristanC: ^do you know | 19:59 |
| *** caphrim007 has quit IRC | 19:59 | |
| tobiash | pabelanger: are you on master or release? | 20:00 |
| pabelanger | tobiash: release | 20:00 |
| pabelanger | so, we need a new tag for nodepool it seems | 20:00 |
| tobiash | That might not yet be released | 20:00 |
| pabelanger | https://zuul-ci.org/docs/nodepool/releasenotes.html#in-development | 20:00 |
| pabelanger | tobiash: yah | 20:00 |
| pabelanger | lets ask Shrews and corvus if we could get a release of nodepool :) | 20:00 |
| corvus | i'd like someone to verify that we're running master in openstack-infra first, otherwise sounds good | 20:02 |
| pabelanger | wfm, I can help do that | 20:02 |
| clarkb | I restarted for some changes prior to ansiblefest iirc | 20:02 |
| clarkb | September 25th | 20:02 |
| clarkb | that change should be in openstack-infra's running launchers | 20:03 |
| clarkb | now to see how far master has moved | 20:03 |
| clarkb | the only changes since then are fixes to tests and zuul configuration | 20:03 |
| pabelanger | yah | 20:04 |
| clarkb | you can probably safely tag master as a result, but can also restart openstack's launchers and double check | 20:04 |
| pabelanger | clarkb: if we want to switch to #openstack-infra, I can help with restarts if we want | 20:05 |
| pabelanger | otherwise, I can see which sha is running | 20:05 |
| tobiash | We should add the version to the web ui so we can check the exact running version easily | 20:05 |
| pabelanger | tobiash: ++ | 20:05 |
| corvus | i'm okay with tagging master without a restart based on the changes clarkb described | 20:08 |
| corvus | let's see if Shrews comes up with a reason not to, otherwise i'll do it in a bit | 20:09 |
| clarkb | corvus: 7015bd9af4bcb8c494cf5f860e2c64e0698de999 is the change I restarted launchers for fwiw | 20:11 |
| openstackgerrit | Merged openstack-infra/zuul master: Do only one status request at a time https://review.openstack.org/609757 | 20:15 |
| pabelanger | okay, thanks. I've just restarted all service for nodepool in openstack-infra. Either commit should work for us in ansible :) | 20:18 |
| *** clarkb has quit IRC | 20:18 | |
| *** clarkb has joined #zuul | 20:18 | |
| ianw | corvus: do you think we could go ahead with https://review.openstack.org/#/c/604980 (attr-overview for options) in zuul-sphinx and get a release of it? this would unblock the nodepool config reorg to be more zuul-ish | 20:27 |
| Shrews | corvus: i have no objections | 20:27 |
| corvus | ianw: thanks for the reminder, i'll look at that as soon as i finish this quickstart stuff | 20:38 |
| *** pcaruana has quit IRC | 20:40 | |
| openstackgerrit | Merged openstack-infra/zuul master: Added steps to install docker for all distros https://review.openstack.org/609765 | 21:26 |
| *** panda has quit IRC | 21:34 | |
| *** panda has joined #zuul | 21:37 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Quick-start: switch to polygerrit https://review.openstack.org/609828 | 21:40 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: WIP: Skeleton code to cleanup down ports https://review.openstack.org/609829 | 21:40 |
| corvus | ianw: lgtm; AJaeger, tobiash: want to take a look at https://review.openstack.org/604980 ? | 21:43 |
| corvus | Shrews, clarkb, pabelanger: based on the release notes, we should probably do 3.3.0. so 3.3.0 @ 32b8f58df0f7764f21f8781749b59bb2d358cd74 (master) sound good? | 21:46 |
| clarkb | corvus: yes new features were added so 3.3.0 sounds good to me. and that sha looks right to me as well | 21:51 |
| *** gouthamr_ is now known as gouthamr | 21:53 | |
| clarkb | corvus: what does noteDb.changes.autoMigrate true in gerrit config do? | 21:54 |
| corvus | clarkb: dunno, it's in the default dockerfile | 21:55 |
| corvus | (i just needed to override the cmd to add the email and polygerrit bits) | 21:55 |
| clarkb | reading the docs I think that is to do online migration for changes into notedb from real db on the 2.13 to 2.14/15 upgrade path | 21:55 |
| corvus | may be unecessary for us then; that might be for folks with longer running containers | 21:56 |
| clarkb | maybe this upgrade won't be so painful for us afterall | 21:56 |
| clarkb | ya, in any case it should be fine to have particularly if upstream has it | 21:56 |
| corvus | (us=zuul quickstart) | 21:56 |
| corvus | switching to infra hat -- i'm really eager to have the new gerrit :) | 21:56 |
| corvus | it's kind of awesome, and polygerrit looks good | 21:57 |
| clarkb | ya polygerrit is basically the old chang e screen made less ugly | 21:57 |
| clarkb | there are a few gerrit 2.15 problem threads on their mailing list I need to read up on, but we should probably start looking at upgrading review-dev soonish | 21:57 |
| corvus | nodepool 3.3.0 pushed (cc pabelanger) | 21:58 |
| clarkb | also single page diffs | 22:00 |
| clarkb | the long desired feature | 22:00 |
| clarkb | details on the OpenStack board meeting being held in berlin are publsihed at https://wiki.openstack.org/wiki/Governance/Foundation/12Nov2018BoardMeeting now | 22:02 |
| clarkb | note the location (if you plan to attend in person) and the rough agenda which is starting to take shape. Morning will be "normal" openstack board meeting, then afternoon is the joint leadership meeting | 22:03 |
| *** openstackgerrit has quit IRC | 22:19 | |
| clarkb | https://review.openstack.org/#/c/609514/8 and its children are showing +1 by zuul now if anyone else wants to review the changes to test the quick start docker stuff | 22:20 |
| clarkb | to clarify the non voting quick start job is also successful | 22:21 |
| *** openstackgerrit has joined #zuul | 22:21 | |
| *** openstackstatus has quit IRC | 22:28 | |
| *** openstackstatus has joined #zuul | 22:30 | |
| *** ChanServ sets mode: +v openstackstatus | 22:30 | |
| clarkb | corvus: ok trying to find the incantation that replaces with_first_found and you are so right about that | 22:32 |
| clarkb | why isn't this a thing builtinto the language anymore | 22:32 |
| ianw | cause i think jinja is turing complete | 22:38 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul-jobs master: Install docker from distro properly https://review.openstack.org/609844 | 22:39 |
| clarkb | corvus: I'm going to edit one of your changes to depends on ^ | 22:40 |
| clarkb | what is zuul/web/static/.keep and why is git so confused about it | 22:42 |
| clarkb | wow I have root owned static dir | 22:43 |
| clarkb | I'm going to guess that the docker method of building the static content does that | 22:43 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul master: Build new container images for quick-start https://review.openstack.org/609763 | 22:45 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul master: Quick-start: switch to polygerrit https://review.openstack.org/609828 | 22:45 |
| clarkb | if I've fixed install-docker correctly ^ should now work | 22:45 |
| pabelanger | corvus: great, thank you | 23:18 |
| pabelanger | clarkb: corvus: actually, looksl like the release jobs failed: http://lists.openstack.org/pipermail/release-job-failures/2018-October/000972.html | 23:21 |
| pabelanger | debugging now | 23:21 |
| clarkb | The task includes an option with an undefined variable. The error was: 'zuul_work_dir' is undefined | 23:22 |
| clarkb | http://logs.openstack.org/90/907251740903ed8c44eb9556d7a34a1e32abd476/release/release-zuul-python/9e1c531/ara-report/result/d317ca20-5f47-4592-925a-c01f7e4e128f/ | 23:22 |
| pabelanger | yah | 23:22 |
| pabelanger | looking to see what changed | 23:22 |
| pabelanger | looks like a new job | 23:23 |
| pabelanger | fixing now | 23:23 |
| clarkb | I've got to pop out to start dinner but will try to review once a fix is up | 23:24 |
| pabelanger | remote: https://review.openstack.org/609855 Fix syntax error with zuul-tarball job | 23:27 |
| pabelanger | should fix it | 23:27 |
| *** rlandy is now known as rlandy|bbl | 23:46 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!