| *** tosky has quit IRC | 00:28 | |
| corvus | clarkb: nope we already do. | 00:29 |
|---|---|---|
| corvus | i was surprised myself; our gerrit is already new enough to require that :) | 00:29 |
| corvus | clarkb: er, i stand corrected | 00:31 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Update sdk to v0.13.0 and install dhall-to-json https://review.opendev.org/702103 | 00:31 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Setup Prelude and Kubernetes binding https://review.opendev.org/702104 | 00:31 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Replace existing operator tasks with the new application https://review.opendev.org/702106 | 00:31 |
| tristanC | so, here is the simplified version ^ it's not as complete as the previous one but it seems to work for the simple case | 00:32 |
| tristanC | basically, the dhall code is now contained in that single file: https://review.opendev.org/#/c/702106/16/conf/zuul/resources.dhall | 00:33 |
| openstackgerrit | Merged zuul/nodepool master: Lower min-ready request priority https://review.opendev.org/704058 | 00:34 |
| tristanC | and there are some test files to run the things without kubernetes ( https://review.opendev.org/#/c/702106/15..16/playbooks/files/cr_spec.yaml ) , or without operator ( https://review.opendev.org/#/c/702106/15..16/playbooks/files/local.yaml ) | 00:40 |
| tristanC | i'll add the missing attributes and rebase the rest of the changes tomorrow. | 00:42 |
| *** johanssone_ has quit IRC | 00:47 | |
| *** johanssone has joined #zuul | 00:52 | |
| *** zxiiro has quit IRC | 01:08 | |
| *** andreaf has quit IRC | 01:37 | |
| *** andreaf has joined #zuul | 01:38 | |
| *** zxiiro has joined #zuul | 01:41 | |
| *** swest has quit IRC | 02:07 | |
| *** rf0lc0 has quit IRC | 02:18 | |
| *** swest has joined #zuul | 02:22 | |
| *** bhavikdbavishi has joined #zuul | 02:55 | |
| *** bhavikdbavishi1 has joined #zuul | 03:02 | |
| *** bhavikdbavishi has quit IRC | 03:04 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:04 | |
| *** bhavikdbavishi has quit IRC | 03:24 | |
| *** bhavikdbavishi has joined #zuul | 03:41 | |
| *** zxiiro has quit IRC | 03:49 | |
| *** rlandy has quit IRC | 04:05 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Replace existing operator tasks with the new dhall function https://review.opendev.org/702106 | 04:50 |
| *** raukadah is now known as chkumar|ruck | 05:00 | |
| *** chkumar|ruck is now known as chkumar|rover | 05:24 | |
| *** swest has quit IRC | 05:26 | |
| *** evrardjp has quit IRC | 05:34 | |
| *** evrardjp has joined #zuul | 05:34 | |
| *** swest has joined #zuul | 06:10 | |
| *** jpena|off is now known as jpena | 08:00 | |
| *** felixedel has joined #zuul | 08:09 | |
| *** at_work has quit IRC | 08:24 | |
| *** notnone has joined #zuul | 08:29 | |
| *** tosky has joined #zuul | 08:36 | |
| openstackgerrit | Merged zuul/zuul-jobs master: fetch-tox-output: introduce zuul_use_fetch_output https://review.opendev.org/681864 | 08:36 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Making opensuse-tumbleweed jobs nv https://review.opendev.org/702693 | 08:39 |
| AJaeger | config-core, zuul-maint, please review ^ to be able to merge some changes in zuul-jobs | 08:40 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Making opensuse-tumbleweed jobs nv https://review.opendev.org/702693 | 08:41 |
| AJaeger | argh, needs a rebase ^ | 08:42 |
| *** evrardjp has quit IRC | 08:54 | |
| *** felixedel has quit IRC | 08:59 | |
| fungi | AJaeger: i take it there's a problem with opendev's tumbleweed mirror again? | 09:01 |
| AJaeger | fungi: yes ;( | 09:02 |
| AJaeger | a file is missing apparently | 09:03 |
| *** pcaruana has joined #zuul | 09:14 | |
| openstackgerrit | Merged zuul/zuul-jobs master: Making opensuse-tumbleweed jobs nv https://review.opendev.org/702693 | 09:31 |
| AJaeger | tristanC: all your zuul_use_fetchoutput changes are in merge-conflict, please rebase and resolve... | 09:35 |
| *** bhavikdbavishi has quit IRC | 09:38 | |
| *** bhavikdbavishi has joined #zuul | 09:39 | |
| *** saneax has joined #zuul | 09:43 | |
| *** saneax has quit IRC | 09:52 | |
| openstackgerrit | Uzume proposed zuul/zuul master: Change default Gerrit HTTP auth method https://review.opendev.org/704068 | 10:00 |
| openstackgerrit | Merged zuul/zuul-jobs master: Return subunit test report as artifact https://review.opendev.org/703235 | 10:04 |
| *** bhavikdbavishi has quit IRC | 10:13 | |
| *** bhavikdbavishi has joined #zuul | 10:14 | |
| *** bhavikdbavishi has quit IRC | 10:18 | |
| *** bhavikdbavishi has joined #zuul | 10:36 | |
| *** bhavikdbavishi has quit IRC | 10:45 | |
| *** avass has joined #zuul | 11:39 | |
| *** rf0lc0 has joined #zuul | 12:14 | |
| *** jpena is now known as jpena|lunch | 12:29 | |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Handle draft pull requests in canMerge https://review.opendev.org/704150 | 12:55 |
| tobiash | pabelanger: you might be interested in ^ to prevent potential gate loops when working with draft pull requests | 12:57 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Handle draft pull requests in canMerge https://review.opendev.org/704150 | 13:04 |
| *** sshnaidm|afk has quit IRC | 13:11 | |
| tristanC | tobiash: it seems like the current zuul-executor image on docker hub (31271b82631a 16 hours ago 1.38 GB) is affected by the `ERROR zuul.ansible_manager: Ansible version 2.9 not working` issue again | 13:15 |
| tristanC | AJaeger: thanks, i'll have a look shortly. | 13:18 |
| *** hashar has joined #zuul | 13:19 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Generate TLS certificats for the gearman service https://review.opendev.org/702716 | 13:25 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Add OpenShift SCC and functional test https://review.opendev.org/702758 | 13:25 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Handle service restart when connections are changed https://review.opendev.org/703624 | 13:25 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Add tenant reconfiguration when main.yaml changed https://review.opendev.org/703631 | 13:25 |
| *** rlandy has joined #zuul | 13:28 | |
| *** rf0lc0 is now known as rfolco|doctor | 13:33 | |
| *** jpena|lunch is now known as jpena | 13:35 | |
| *** jamesmcarthur has joined #zuul | 13:48 | |
| openstackgerrit | Antoine Musso proposed zuul/zuul master: On CI, use a tmpfs for ZUUL_TEST_ROOT https://review.opendev.org/704160 | 14:10 |
| *** sshnaidm has joined #zuul | 14:13 | |
| *** sshnaidm has quit IRC | 14:17 | |
| openstackgerrit | Antoine Musso proposed zuul/zuul master: On CI, use a tmpfs for ZUUL_TEST_ROOT https://review.opendev.org/704160 | 14:23 |
| *** sshnaidm has joined #zuul | 14:24 | |
| *** zxiiro has joined #zuul | 14:24 | |
| pabelanger | tobiash: acK! | 14:25 |
| *** hashar has quit IRC | 14:29 | |
| *** sshnaidm_ has joined #zuul | 14:31 | |
| *** sshnaidm has quit IRC | 14:31 | |
| fungi | i'm trying to come up with missing terms for our zuul glossary, since it's very sparse at the moment. here's what i've got so far i could add: artifact, connection, job, job variant, node, parent job, pipeline, project, project gating, project template | 14:40 |
| fungi | anything else anyone think should be added? | 14:40 |
| fungi | oh, tenant should probably be in there too | 14:40 |
| tristanC | the executor image issue can be observed here: https://zuul.opendev.org/t/zuul/build/e7b91711195d418fb6f3f9d0dc2a2fbf/log/docker/k8s_executor_zuul-executor-0_myproject_2b25e25c-3eb2-11ea-998a-fa163e6bfc57_5.txt#26-44 | 14:41 |
| dmsimard | fungi: speculative merging ? gating ? depends-on ? | 14:44 |
| *** panda is now known as panda|bbl | 14:45 | |
| fungi | gating could be used instead of project gating, i guess | 14:45 |
| fungi | but we already have gate in there for a related purpose | 14:45 |
| dmsimard | ah, yeah, project gating is fine -- had misread | 14:46 |
| fungi | depends-on is a commit message directive not a term, i don't think, but we could maybe have an entry on cross-project dependencies | 14:46 |
| fungi | i wonder if speculative execution would be a better term than speculative merging. i'll see if we say either anywhere in the other docs | 14:47 |
| fungi | oh, build and buildset should get entries | 14:48 |
| dmsimard | nodeset is also missing | 15:00 |
| *** hashar has joined #zuul | 15:03 | |
| openstackgerrit | Antoine Musso proposed zuul/zuul master: On CI, use a tmpfs for ZUUL_TEST_ROOT https://review.opendev.org/704160 | 15:04 |
| openstackgerrit | Antoine Musso proposed zuul/zuul master: On CI, use a tmpfs for ZUUL_TEST_ROOT https://review.opendev.org/704160 | 15:05 |
| *** jamesmcarthur has quit IRC | 15:08 | |
| fungi | dmsimard: ooh, yep good suggestion | 15:12 |
| corvus | fungi: ++ | 15:12 |
| *** ChrisShort has quit IRC | 15:18 | |
| *** ChrisShort has joined #zuul | 15:18 | |
| *** chkumar|rover is now known as raukadah | 15:19 | |
| *** sshnaidm_ is now known as sshnaidm|off | 15:25 | |
| *** jamesmcarthur has joined #zuul | 15:28 | |
| *** mnasiadka has quit IRC | 15:31 | |
| *** mnasiadka has joined #zuul | 15:32 | |
| hashar | hi :) Iam throwing an idea out of the blue. zuul-build-image builds the Docker images and gets paused to keep the docker registry available | 15:36 |
| hashar | which would mean that jobs keeps a node busy in Nodepool while other jobs in the build set apparently do not need the registry to be available | 15:36 |
| corvus | hashar: it's used by zuul-quick-start | 15:38 |
| hashar | ahhh | 15:38 |
| hashar | that is the dependency I was looking ;] | 15:38 |
| corvus | yeah, i'd like to find a way to speed that up a bit. it's supposed to get a node for the second job with high priority, but that doesn't happen as quickly as i would like. | 15:39 |
| hashar | the job is paused for the duration of the whole buildset isn't it? | 15:39 |
| corvus | hashar: no, just until its dependencies are done | 15:39 |
| corvus | (but because of the dependency chain, it may be the last to finish) | 15:39 |
| hashar | I just noticed zuul-quick-start has a "requires: docker-image" which is I guess how the dependency is defined | 15:40 |
| hashar | thx ;) | 15:41 |
| corvus | hashar: no, that's actually artifact transfer from other buildsets; the dependency is set with "dependencies" on the job | 15:41 |
| corvus | hashar: https://opendev.org/zuul/zuul/src/branch/master/.zuul.yaml#L227-L228 | 15:41 |
| hashar | neat | 15:42 |
| corvus | hashar: the "requires" thing means that if there is a change ahead of it in the pipeline that "provides: docker-image", zuul-build-image will wait for that job to complete before it starts | 15:42 |
| corvus | (and it will use the built image from the one ahead) | 15:43 |
| corvus | this lets us do things like make changes to base images and then use them in child images | 15:43 |
| corvus | (eg, zuul images are "FROM opendev/python-base", so we can make a change to python-base and then see that reflected in zuul's image) | 15:44 |
| hashar | ahhhhh | 15:44 |
| hashar | that might be a little bit too much of advanced use case for me right now ;] | 15:45 |
| corvus | yeah, it's not necessarily for everyone, but like most of zuul, when you need it, you need it. :) | 15:45 |
| hashar | I am going to dig into ansible, I got to learn it eventually | 15:46 |
| hashar | and as an excuse to learn, I will try to speed up the zuul test suite | 15:46 |
| *** samccann has quit IRC | 15:47 | |
| *** samccann has joined #zuul | 15:47 | |
| *** panda|bbl is now known as panda | 16:02 | |
| openstackgerrit | Merged zuul/zuul master: Fix canceling builds in starting phase https://review.opendev.org/684414 | 16:06 |
| *** tosky has quit IRC | 16:16 | |
| tobiash | \o/ | 16:17 |
| *** mauved has quit IRC | 16:20 | |
| *** mauved has joined #zuul | 16:21 | |
| *** hashar has quit IRC | 16:28 | |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Update component diagram for GitHub / Web events https://review.opendev.org/704172 | 16:37 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Update component diagram for GitHub / Web events https://review.opendev.org/704172 | 16:41 |
| *** rfolco|doctor is now known as rfolco | 16:45 | |
| *** Goneri has joined #zuul | 16:50 | |
| clarkb | hashar isn't here anymore but I'm trying to quantify if https://review.opendev.org/704160 is faster right now | 16:56 |
| corvus | clarkb: my recollection is that it didn't seem to make much of a difference on the cloud providers available the last time we tried it several years ago | 16:58 |
| corvus | probably due to nuances of how storage worked on them | 16:58 |
| corvus | it *certainly* makes a difference for me locally though. so i'm +2 on the principle of it being a good example to follow and probably not hurting in the gate. | 16:59 |
| corvus | (also, i don't want to wear out my ssd) | 17:00 |
| clarkb | ya it doesn't seem to make anything worse at least | 17:00 |
| *** hashar has joined #zuul | 17:05 | |
| clarkb | with a very low sample count of one before and one after on the same clouds py35 is ~30 seconds quicker and py37 is ~400 seconds quicker | 17:05 |
| clarkb | depending on the cloud this may actually represent a nice improvement (rax-ord had the big increase, inap-mtl01 the small one) | 17:06 |
| corvus | cool, that sounds promising then | 17:06 |
| *** hashar has quit IRC | 17:06 | |
| corvus | hashar: ^ this is re your tmpfs change; you missed a few lines (you can check eavesdrop to catch up but probably not that important :) | 17:06 |
| openstackgerrit | Merged zuul/zuul-jobs master: Update roles/prepare-workspace-openshift/README.rst https://review.opendev.org/704017 | 17:08 |
| clarkb | https://zuul.opendev.org/t/zuul/build/9e6f918fdc8d49dfa34a3603e1d39553/log/job-output.txt#374 shows it is working too | 17:08 |
| clarkb | with that I'll approve | 17:09 |
| *** jpena is now known as jpena|off | 17:11 | |
| corvus | tobiash: https://review.opendev.org/691715 has been sitting with a +2 for a while if you don't mind taking a look | 17:11 |
| tobiash | corvus: oops, I've overlooked that one, I'll review in a bit | 17:30 |
| openstackgerrit | Merged zuul/zuul master: test: prevent ResourceWarning in test_client https://review.opendev.org/703782 | 17:32 |
| tobiash | corvus: lgtm | 17:38 |
| *** rlandy is now known as rlandy|brb | 17:41 | |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Add estimated time remaining tooltip to UI https://review.opendev.org/703892 | 17:42 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: WIP: debug estimate_time on job start https://review.opendev.org/704179 | 17:42 |
| openstackgerrit | Merged zuul/zuul master: Whitelist debug threads used by visual studio code https://review.opendev.org/703248 | 17:49 |
| *** jamesmcarthur has quit IRC | 17:52 | |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Use apt mirror infrastructure during zuul-quick-start https://review.opendev.org/649448 | 17:56 |
| tobiash | corvus, tristanC: re manage-ansible. What do you think if we make ansible installation single threaded by default (for the container build) and optionally multi threaded combined with different cache dirs per thread? | 18:01 |
| tobiash | that way we could combine a reliable installation during the container build and still having faster setup times for the tox runs | 18:01 |
| clarkb | tobiash: different cache dirs should fix the contention problem with the packages and wheels right? | 18:01 |
| tobiash | clarkb: I think so | 18:01 |
| clarkb | I guess if that works in the tox jobs then we can switch the container builds to it? | 18:02 |
| tobiash | at least that's my assumption based on the various errors I've seen there | 18:02 |
| tobiash | clarkb: the container builds need to share a single cache dir because of the build process | 18:02 |
| clarkb | oh right | 18:02 |
| clarkb | we stage the packages to avoid needing buidl deps in the runtime | 18:03 |
| tobiash | yes | 18:03 |
| tobiash | ok, we could get around this by manually merging the cache dirs afterwards | 18:03 |
| tobiash | oh actually, we don't need the cache from zuul-manage-ansible at all | 18:05 |
| tobiash | we just copy the ansible envs into the dest image | 18:05 |
| tobiash | so yes, if it works in tox we could leverage this in the container build as well | 18:05 |
| tristanC | not sure i understood what is going on, is this a pip issue? | 18:05 |
| tobiash | tristanC: the issue is that multiple pip processes work on the same cache while installing ansible into different venvs | 18:06 |
| tobiash | so the solution to keep multi-process installation would be to tell pip to use individual cache dirs | 18:07 |
| tristanC | tobiash: iiuc, if pip can't run concurrently, then this a pip bug where it should use a lock file to prevent that install failure? | 18:09 |
| *** rlandy|brb is now known as rlandy | 18:09 | |
| tobiash | yes, probably | 18:09 |
| tristanC | it seems already reported https://github.com/pypa/pip/issues/2361 | 18:11 |
| tobiash | this one is about concurrently installing packages into the same environment, which is actually harder | 18:12 |
| tobiash | our issue is concurrently installing packages into different environments | 18:13 |
| tobiash | I found this one (which seems to be fixed since ages): https://github.com/pypa/pip/issues/1141 | 18:16 |
| tobiash | but maybe the wheel cache isn't thread safe as well | 18:16 |
| tobiash | or it got broken again in the meantime | 18:18 |
| tobiash | nevertheless I think we probably don't want to rely on thread safety of pip so we can either go single threaded or with individual cache dirs or both | 18:19 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: WIP: debug estimate_time on job start https://review.opendev.org/704179 | 18:20 |
| *** jamesmcarthur has joined #zuul | 18:24 | |
| tobiash | pabelanger: did you already try the js tooling for hacking on zuul-web? (asking because of ^) | 18:25 |
| zbr | can we either merge https://review.opendev.org/#/c/702473/ or increase the timeouout from 3600 to 4000 ? | 18:25 |
| zbr | https://zuul.opendev.org/t/zuul/builds?job_name=tox-py35&project=zuul%2Fzuul | 18:26 |
| tristanC | tobiash: if that issue can be solved/mitigated by pip/wheel, it seems safer to go single threaded until it's fixed, otherwise i can't tell what is better (faster but more i/o intensive, or slower but using a single cache...) | 18:26 |
| tobiash | zbr: as there was opposition to 702473 we should increase the timeout, do you want to propose that? | 18:27 |
| zbr | sure, doing it now | 18:27 |
| tobiash | thx :) | 18:27 |
| tobiash | tristanC: well, separated caches are safe as well (we also can have both ways if we want) | 18:27 |
| tristanC | tobiash: fwiw, we have been able to install all the ansible using a single venv, it seems like 2.[6789] works with the same requirements.txt, we only keep the ansible modules and cli separated | 18:28 |
| tristanC | though we use a special rpm spec script to do that, i don't know if venv can inherit from a non system-wide venv | 18:29 |
| tristanC | tobiash: is there a patch for separated caches? otherwise i can have look now if you want | 18:30 |
| tobiash | tristanC: right now it's just an idea :) | 18:30 |
| tobiash | we could also just disable the cache instead of putting it into a tmpdir and throwing it away: --no-cache-dir | 18:31 |
| tobiash | pip supports that as well | 18:31 |
| tobiash | if that works, that would be easy | 18:31 |
| tobiash | and I also checked, we don't need that cache in the container builds | 18:32 |
| zbr | anyone here against yaml anchors in job files? | 18:35 |
| clarkb | I think we use them for file matchers/excluders | 18:35 |
| clarkb | should befine | 18:35 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Disable pip cache when managing ansible https://review.opendev.org/704184 | 18:35 |
| tobiash | tristanC: ^ | 18:35 |
| tristanC | why would anyone be against yaml anchors? :) | 18:36 |
| zbr | they are very useful to replicate stuck from check to gate | 18:37 |
| openstackgerrit | Sorin Sbarnea proposed zuul/zuul master: Increase timeout for tox-py27/36 jobs https://review.opendev.org/704185 | 18:38 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: DNM: Test multi-threaded ansible setup without cache https://review.opendev.org/704186 | 18:39 |
| clarkb | zbr: one small nit on the commit message | 18:39 |
| openstackgerrit | Sorin Sbarnea proposed zuul/zuul master: Increase timeout for tox-py35/37 jobs https://review.opendev.org/704185 | 18:41 |
| tobiash | zbr: this conflicts with the first change in gate, you might want to rebase ^ | 18:44 |
| tobiash | (in around 10min) | 18:45 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Add estimated time remaining tooltip to UI https://review.opendev.org/703892 | 18:56 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: DNM: test arbritary uid support for zuul image https://review.opendev.org/702980 | 18:57 |
| clarkb | thinking out loud here about job runtimes and how setup is a significant portion of that runtime. I seem to recall that ansible at some point started forking for every task and that slowed ansible down quite a bit | 18:58 |
| clarkb | do any of the ansible folks know if there are any plans to go back to the old pre forking behavior? | 18:59 |
| openstackgerrit | Merged zuul/zuul master: On CI, use a tmpfs for ZUUL_TEST_ROOT https://review.opendev.org/704160 | 19:01 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-operator master: Add zuul-operator-promote-image job https://review.opendev.org/704187 | 19:04 |
| tristanC | tobiash: perhaps the image is correct, it seems like for some reason, it's only the ansible-2.9 venv that failed to be validated when running in openshift... e.g. https://review.opendev.org/#/c/703631/3 and it's git-parent . Maybe https://review.opendev.org/#/c/702980/ will fix that issue | 19:06 |
| tristanC | tobiash: note that i'm not using the privileged scc, so the pod are still running with the AnyUid policy | 19:08 |
| pabelanger | tristanC: clarkb: fungi: https://review.opendev.org/703892 should address comments from yesterday now | 19:08 |
| clarkb | pabelanger: cool, its the null check that we needed? | 19:08 |
| pabelanger | yah | 19:08 |
| clarkb | I'll verify it against the preview site and vote after | 19:08 |
| pabelanger | we sould also say, job preparing to run on null | 19:09 |
| pabelanger | but this works | 19:09 |
| tristanC | pabelanger: nice, thanks! i'll check the '' title doesn't yield a empty tooltip, otherwise lgtm | 19:11 |
| openstackgerrit | Merged zuul/zuul master: Don't add implicit role if plugin found https://review.opendev.org/691715 | 19:11 |
| pabelanger | tristanC: ++ I didn't see it in chrome | 19:12 |
| pabelanger | also, happy to update logic with assistance | 19:12 |
| pabelanger | I wish it said 2 hours and 15 mins, but meh, this is better then nothing | 19:12 |
| clarkb | could you do hours:minutes:seconds remaining? | 19:13 |
| clarkb | its a bit of math but not too bad | 19:14 |
| pabelanger | yes, but I think if time was great then 24 hours, it would become 0 hours | 19:14 |
| clarkb | well you could do days too | 19:15 |
| pabelanger | https://stackoverflow.com/a/41453922 | 19:15 |
| tristanC | pabelanger: aren't those estimate anyway? | 19:16 |
| clarkb | oh I was thinking of doing the math directly | 19:16 |
| pabelanger | tristanC: yup, estimate but sometime good to know if closer to 2 hours then 3 hours :) | 19:17 |
| pabelanger | 1 hour window is better then nothing right now | 19:17 |
| clarkb | MINUTE=60 ; HOUR=MINUTE*60; DAY=24*HOUR; days=time/DAY ; remainer=time%DAY ; hours=remainder/HOUR ; remainder=remainder%HOUR; minutes=remainder/MINUTE ; seconds=remainder%MINUTE | 19:18 |
| clarkb | pabelanger: ^ that is the math | 19:19 |
| clarkb | assuming time is whole seconds | 19:19 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Increase timeout for tox-py35/37 jobs https://review.opendev.org/704185 | 19:19 |
| clarkb | pabelanger: you could also simplify it and calculate days first then feed the remainder into your format function | 19:19 |
| pabelanger | agree, maybe for next friday :) | 19:20 |
| *** jamesmcarthur has quit IRC | 19:23 | |
| openstackgerrit | Merged zuul/zuul master: Update component diagram for GitHub / Web events https://review.opendev.org/704172 | 19:48 |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Be more specific with remaining time https://review.opendev.org/704191 | 19:55 |
| clarkb | pabelanger: ^ thats a follow on with the above formula | 19:55 |
| clarkb | it may need tweaking to get right (I am not 100% sure about js maths) | 19:55 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul master: Dockerfile: add support for arbritary uid https://review.opendev.org/650246 | 19:56 |
| tristanC | zuul-maint : could you please review https://review.opendev.org/#/c/650246/6 , it already has a +2 and I added links to a job being fixed by the change in the review comment | 20:00 |
| openstackgerrit | Merged zuul/zuul master: Increase timeout for tox-py35/37 jobs https://review.opendev.org/704185 | 20:16 |
| openstackgerrit | Tristan Cacqueray proposed zuul/nodepool master: kubernetes: add release note about service account token issue https://review.opendev.org/703869 | 20:20 |
| *** hashar has joined #zuul | 20:30 | |
| *** hashar has quit IRC | 20:41 | |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Be more specific with remaining time https://review.opendev.org/704191 | 20:55 |
| *** hashar has joined #zuul | 20:58 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: fetch-subunit-output: introduce zuul_use_fetch_output https://review.opendev.org/681882 | 21:05 |
| openstackgerrit | Antoine Musso proposed zuul/zuul master: Add client_id to RPC client https://review.opendev.org/704201 | 21:07 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: fetch-subunit-output: introduce zuul_use_fetch_output https://review.opendev.org/681882 | 21:08 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: DNM: test tox-py36 on openshift node https://review.opendev.org/682049 | 21:08 |
| *** avass has quit IRC | 21:15 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: fetch-javascript-tarball: introduce zuul_use_fetch_output https://review.opendev.org/681603 | 21:15 |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Be more specific with remaining time https://review.opendev.org/704191 | 21:18 |
| *** zxiiro has quit IRC | 21:23 | |
| *** rlandy has quit IRC | 21:44 | |
| pabelanger | Friday afternoon questions. By using https://zuul-ci.org/docs/zuul/reference/jobs.html#pausing-the-job and starting child jobs, we might have a want to unpause the parent before the child finished. Mostly because child jobs take upwards of 2.5 hours, and we only need the parent for first 15mins of the job | 21:54 |
| *** rfolco has quit IRC | 22:00 | |
| *** armstrongs has joined #zuul | 22:15 | |
| *** jamesmcarthur has joined #zuul | 22:21 | |
| *** armstrongs has quit IRC | 22:24 | |
| *** hashar has quit IRC | 22:49 | |
| *** jamesmcarthur has quit IRC | 22:57 | |
| clarkb | pabelanger: I think the trick would be formalizing the signal that all children that want to access those resources are finished with them | 23:00 |
| fungi | do we have a mechanism for a build to communicate any status change back to the scheduler before completion? | 23:08 |
| clarkb | fungi: zuul_return is per playbook I think so could probably be mangled to do this | 23:09 |
| clarkb | but not 100% sure on that | 23:09 |
| fungi | we did at one point talk about making it possible for a build to inform the scheduler early when it knows it's going to fail, so that pipeline changes can take effect sooner | 23:10 |
| fungi | maybe this falls into a similar category | 23:13 |
| *** jamesmcarthur has joined #zuul | 23:14 | |
| *** jamesmcarthur has quit IRC | 23:23 | |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Be more specific with remaining time https://review.opendev.org/704191 | 23:32 |
| *** jamesmcarthur has joined #zuul | 23:40 | |
| clarkb | hrm seems like zuul-tox-remote might be flaky | 23:50 |
| clarkb | tests.remote.test_remote_zuul_stream.TestZuulStream29.test_command is timing out? | 23:51 |
| clarkb | https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_ac4/703892/7/gate/zuul-tox-remote/ac40b75/job-output.txt is an example | 23:51 |
| pabelanger | clarkb: fungi: yah, I was thinking each child job would use zuul_return to signal back some how to unpause, then once all child have done so the parent would resume and clean up | 23:57 |
| clarkb | pabelanger: but not all children may need the parent | 23:58 |
| clarkb | I guess it should't parent in that case | 23:58 |
| clarkb | so ya that could be the implication | 23:58 |
| pabelanger | yah | 23:59 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!