| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Remove state_dir requirement from merger https://review.openstack.org/490682 | 00:11 |
|---|---|---|
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat github driver docs https://review.openstack.org/490688 | 00:34 |
| *** http_GK1wmSU has joined #zuul | 02:34 | |
| *** http_GK1wmSU has left #zuul | 02:35 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Prevent execution of locally overridden core modules https://review.openstack.org/490216 | 03:21 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Add some ansible plugin tests https://review.openstack.org/454396 | 03:21 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Allow and document use of the uri module from localhost https://review.openstack.org/490215 | 03:21 |
| mordred | rebased to fix merge conflict caused by one of the docs patches | 03:22 |
| mordred | jeblair: it might be? | 03:22 |
| mordred | jeblair, SpamapS, pabelanger: https://review.openstack.org/454396 passes tests now if you have a chance to poke it in | 03:27 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat merger and executor config docs https://review.openstack.org/490665 | 03:31 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat web section https://review.openstack.org/490666 | 03:32 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: update format in connections.rst https://review.openstack.org/490672 | 03:32 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat gerrit driver docs https://review.openstack.org/490681 | 03:33 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Remove state_dir requirement from merger https://review.openstack.org/490682 | 03:34 |
| *** bhavik1 has joined #zuul | 04:28 | |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Add some ansible plugin tests https://review.openstack.org/454396 | 05:12 |
| *** bhavik1 has quit IRC | 07:01 | |
| *** amoralej|off is now known as amoralej | 07:23 | |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/nodepool master: Clarify diskimage names in docs https://review.openstack.org/490781 | 07:42 |
| *** http_GK1wmSU has joined #zuul | 08:03 | |
| *** http_GK1wmSU has left #zuul | 08:04 | |
| *** electrofelix has joined #zuul | 08:42 | |
| *** openstackgerrit has quit IRC | 10:18 | |
| *** jkilpatr has joined #zuul | 11:01 | |
| *** dkranz_ has joined #zuul | 12:22 | |
| electrofelix | I've notice an odd behaviour when testing with the bonnyci fork of zuul for github support (as zuulv3 is not recommended for production usage) and it seems like it might be something unrelated to github and a limitation of zuulv2 | 12:35 |
| electrofelix | if you have two pipelines with the same trigger definition but different status reporting sections, and create two project definitions and assign each to one of the two pipelines | 12:36 |
| electrofelix | when an event occurs it appears that both pipelines receive the event and attempt to report the status instead of only the pipeline of the project the event should have been associated with | 12:36 |
| *** amoralej is now known as amoralej|lunch | 12:43 | |
| electrofelix | I recognize that having two pipelines receive the same event but for different projects is probably an edge case | 12:46 |
| electrofelix | but it seems useful to be able to add a test pipeline with updated settings for a single project before updating all other projects with the same setting change | 12:46 |
| *** jkilpatr has quit IRC | 12:49 | |
| tobiash | electrofelix: for the v2 branch I had a similar problem which is fixed by https://review.openstack.org/#/c/334378/ | 13:00 |
| tobiash | electrofelix: you coult try that patch in your environment and see if it helps (not sure if it's the same root cause) | 13:00 |
| *** jkilpatr has joined #zuul | 13:05 | |
| electrofelix | tobiash: thanks, it should clean up my unit test if it does | 13:15 |
| electrofelix | tobiash: confirmed that is the fix | 13:21 |
| *** openstackgerrit has joined #zuul | 13:24 | |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat github driver docs https://review.openstack.org/490688 | 13:24 |
| *** amoralej|lunch is now known as amoralej | 13:24 | |
| *** xinliang has quit IRC | 13:27 | |
| pabelanger | morning | 13:29 |
| *** xinliang has joined #zuul | 13:40 | |
| *** xinliang has quit IRC | 13:40 | |
| *** xinliang has joined #zuul | 13:40 | |
| *** jkilpatr_ has joined #zuul | 13:41 | |
| *** jkilpatr has quit IRC | 13:42 | |
| tobiash | electrofelix: yay | 13:52 |
| mordred | electrofelix: heya - I'd use v3, not that branch | 14:41 |
| mordred | electrofelix: we're rolling v3 out for openstack in roughly 4 weeks - while it's possible we might still break one or two small things between now and then, those breaks will be WAY less of a migration than migrating from fork-of-v2 -> v3 | 14:43 |
| pabelanger | ++ | 14:47 |
| mordred | electrofelix: that fork of v2 is _definitely_ not recommeneded for production at all and we're close enough to prime time that bug reports on v3 or finding issues either in the code or the docs is massively helpful | 14:49 |
| electrofelix | mordred: unfortunately I expect we're committed to being stuck on it for a few more weeks at least, though it sounds like we might be ready to migrate to the v3 branch around the same time it goes live | 14:51 |
| *** jasondotstar has quit IRC | 15:03 | |
| tobiash | mordred: what are the plans for the master branch once zuulv3 is released? | 15:10 |
| tobiash | mordred: asking because I still have some patches for master around and thinking if I should abandon these or keep them for some eventually to be created v2-legacy branch | 15:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Support longer pkcs1-oaep secrets https://review.openstack.org/490620 | 15:12 |
| jeblair | tobiash: our plans are to merge v3 into master; we don't intend to maintain v2 after v3 is released | 15:14 |
| jeblair | tobiash: (i expect the actual v3 release will come from the master branch) | 15:14 |
| tobiash | jeblair: that's clear and I also don't expect to maintain v2 | 15:15 |
| tobiash | jeblair: so I'll just abandon these patches? (currently they're just visual noise on my dashboard) | 15:15 |
| jeblair | tobiash: unless they're critical bugfixes (especially ones that would still affect v3) | 15:16 |
| tobiash | jeblair: https://review.openstack.org/#/q/status:open+project:openstack-infra/zuul+branch:master+owner:%22Tobias+Henkel+%253Ctobias.henkel%2540bmw.de%253E%22 | 15:17 |
| tobiash | not really critical bugfixes | 15:17 |
| tobiash | I'll probably keep 482856 in case gerrit gets updated before zuul | 15:18 |
| clarkb | tobiash: I think we want to be able to point at that one and tell people to apply it locally if they need to talk to newer gerrit with old zuul (so keeping it open is good) | 15:20 |
| jeblair | ++ | 15:21 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Don't request empty nodesets https://review.openstack.org/487243 | 15:21 |
| tobiash | jeblair: what about 403559 (that is a bugfix for zuul-cloner)? | 15:22 |
| tobiash | that's the last one missing in question | 15:22 |
| jeblair | tobiash: i'd personally rather focus on v3 and not take the (admittedly, very small) risk, but if anyone else wants to +3 it that's fine | 15:27 |
| tobiash | jeblair: ok, then I'll wait until monday and abandon it then if no one wants to +3 it :) | 15:28 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: DNM - testing base-test playbook https://review.openstack.org/490945 | 15:33 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat SMTP driver docs https://review.openstack.org/490947 | 15:37 |
| pabelanger | jeblair: mordred: going to restart zuul-executor to pickup new zuul.conf changes | 15:38 |
| pabelanger | I don't see a reload option ATM | 15:38 |
| pabelanger | Ya, I am guessing we'll want a reconfigure_handler for zuul-executor too | 15:39 |
| pabelanger | can see about adding that today | 15:39 |
| jeblair | mordred: what do you want to do about the question in 488221? I'd love to land that | 15:42 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat sql driver docs https://review.openstack.org/490950 | 15:44 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat timer driver docs https://review.openstack.org/490951 | 15:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat timer driver docs https://review.openstack.org/490951 | 15:47 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat zuul driver docs https://review.openstack.org/490953 | 15:49 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul feature/zuulv3: Fix job example and misspelling https://review.openstack.org/490955 | 15:56 |
| Shrews | very simple +3 for someone ^^^ | 15:56 |
| Shrews | that 'image' to 'label' change had me scratching my head for a bit | 15:56 |
| tobiash | Shrews: +3'd | 15:58 |
| mordred | jeblair: oh - sorry - I forgot about that ... one sec | 16:00 |
| mordred | jeblair: I think lets' just do 255 for now - it's a varchar field anyway so there's not a TON of importance in limiting it further | 16:00 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Update SQL reporter to store results https://review.openstack.org/488221 | 16:01 |
| mordred | jeblair: ^^ | 16:01 |
| jeblair | mordred: cool | 16:03 |
| jeblair | tobiash: want to +3 that ^ | 16:03 |
| tobiash | jeblair: nope | 16:05 |
| Shrews | mordred: I'm experimenting with job configuration using shade (e.g., https://review.openstack.org/490948). Something weird happened with the log streaming. Here is what I captured from the websocket stream (http://paste.openstack.org/show/617553/), but what was reported to gerrit was "test-job finger://ze01.openstack.org/3e85265630c94203bb44829314950d45 : FAILURE in 32s" with no link to the log | 16:05 |
| tobiash | upgrade script seems not to fit to code now (16 bs 255 chars) | 16:05 |
| jeblair | tobiash: thanks :) | 16:06 |
| jeblair | mordred: ^ | 16:06 |
| * Shrews acting as the pre-PTG confused guinea pig | 16:06 | |
| tobiash | jeblair: have a comment/question in 490953 | 16:07 |
| jeblair | Shrews: the issue is that job does not inherit from the base job, which is what is responsible for uploading logs | 16:07 |
| Shrews | jeblair: ah! well that explains that | 16:07 |
| jeblair | Shrews: next week i plan on working on story 2001110 | 16:07 |
| jeblair | Shrews: which is to make base jobs both required and implicit | 16:08 |
| jeblair | Shrews: so, if things go well with that, if you had written that job at the end of next week, it should have worked like you intended :) | 16:08 |
| tobiash | Shrews: I encountered this error everytime I had an ansible error in a task such as undefined variable | 16:08 |
| Shrews | jeblair: this is good info | 16:08 |
| mordred | tobiash: DOH | 16:08 |
| jeblair | Shrews: (ie, if you omit 'parent', it will automatically default to 'parent: base' -- which is what you want to add to that to fix it today) | 16:08 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Update SQL reporter to store results https://review.openstack.org/488221 | 16:09 |
| mordred | tobiash: thank you for that catch | 16:09 |
| Shrews | jeblair: that explains the log issue. still wondering why the command failed, but that helps. thx | 16:09 |
| tobiash | mordred: so now all 255 and +2 :) | 16:10 |
| Shrews | i'm guessing 'tox' isn't installed since i bypass our std jobs | 16:10 |
| mordred | Shrews: yes. tox isn't instlaled - and also the base job handles getting the source code over there | 16:11 |
| jeblair | Shrews, mordred: yeah, that's a good reason for it not to work, but it's not a good reason for us not to see any output... | 16:11 |
| mordred | Shrews: and also the cwd for the playbook is /home/zuul but the source code (once the base job copies it over) is in /home/zuul/src/git.openstack.org/openstack-infra/shade | 16:11 |
| mordred | Shrews: aka "src/{{ zuul.project.canonical_name }}" | 16:12 |
| mordred | Shrews: so for things you want to happen "in the source dir of the one project" - you need to add an args:\n cwd: "src/{{ zuul.project.canonical_name }}" to the command | 16:13 |
| Shrews | mordred: ah. i expect this to be a common issue | 16:13 |
| mordred | jeblair: ^^ I don't have an answer yet, but I feel like we may want to figure out a cantrip of some sort to make that easier for the simple case | 16:13 |
| mordred | Shrews: yah. | 16:14 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Fix job example and misspelling https://review.openstack.org/490955 | 16:15 |
| jeblair | mordred: some ideas: a new variable so that's easier to type, or a new command module. | 16:15 |
| Shrews | mordred: jeblair: ok, once i set parent, i get a better log: http://logs.openstack.org/48/490948/4/check/test-job/40d0d7e/job-output.txt.gz#_2017-08-04_16_12_21_812873 | 16:15 |
| jeblair | mordred: are you troubled by the lack of output for the tox command error? | 16:15 |
| pabelanger | Shrews: yay | 16:16 |
| Shrews | curious that didn't show up in the previous run | 16:16 |
| jeblair | Shrews: assuming you didn't change anything else, i note that the old run had "[Zuul] Log Stream did not terminate", but the new one had an actual error | 16:16 |
| Shrews | jeblair: all i did was add the parent | 16:16 |
| jeblair | mordred, Shrews: could there be a race in the command output streaming? | 16:16 |
| * Shrews apologizes if he's causing noise around this | 16:17 | |
| mordred | jeblair: our base job starts the log streamer | 16:17 |
| jeblair | mordred: ah that would do it :) | 16:17 |
| Shrews | ah, yeah. requiring the base job seems even more important now :) | 16:18 |
| jeblair | test-job ERROR No valid playbook found | 16:18 |
| mordred | I am still MILDLY concerned, as I thought our command module still put stdout/stderr into the stdout_lines in the result object | 16:18 |
| jeblair | neat ^ :) had only seen that in tests up to this point. :) | 16:19 |
| mordred | jeblair: neat! | 16:19 |
| jeblair | mordred: i thought you filtered that out for console display, and only included it in json now? | 16:19 |
| jeblair | (since it's presumably redundant when we actually have console streaming) | 16:19 |
| pabelanger | mordred: jeblair: restarting ze01.o.o now that we have no running jobs | 16:19 |
| pabelanger | and started again | 16:21 |
| Shrews | yay, it's working! thx all. sorry to distract | 16:24 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Use zuul.executor.work_root for add-build-sshkey https://review.openstack.org/490969 | 16:29 |
| jeblair | Shrews: no problem! | 16:30 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat tenant config docs https://review.openstack.org/490970 | 16:30 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Update SQL reporter to store results https://review.openstack.org/488221 | 16:32 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Remove score attribute from sqlreporter config https://review.openstack.org/490972 | 16:39 |
| jeblair | mordred, tobiash: ^ oops we forgot something | 16:39 |
| jeblair | also, that's our first reporter with no configuration whatsoever. i think "foo: {}" is the best way to do that, but it's worth a second thought. | 16:39 |
| tobiash | jeblair: would just 'foo:' also work? | 16:44 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Remove configure-mirrors from unittests/pre https://review.openstack.org/490977 | 16:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat timer driver docs https://review.openstack.org/490951 | 16:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat sql driver docs https://review.openstack.org/490950 | 16:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat SMTP driver docs https://review.openstack.org/490947 | 16:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat tenant config docs https://review.openstack.org/490970 | 16:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat zuul driver docs https://review.openstack.org/490953 | 16:46 |
| *** electrofelix has quit IRC | 16:47 | |
| jeblair | tobiash: hrm i'll try | 16:47 |
| tobiash | jeblair: I think that should give you a dict key with none as value | 16:47 |
| pabelanger | jeblair: mordred: so, I think we might need to bikeshed over 490997 ^. It raises another question, do we expect zuul operators to setup local site-variables to use playbooks / roles in zuul-jobs, and if so, we'll need to establish a namespace for such variables I believe | 16:48 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Use null values instead of the empty dict for sql reporter https://review.openstack.org/490978 | 16:49 |
| jeblair | tobiash: yes i think you're correct ^ that's a followup change that does that | 16:49 |
| jeblair | tobiash: i guess that's a little more ansible like from what i've seen | 16:49 |
| jeblair | pabelanger: yes i think we do expect zuul ops to set up local site vars for zuul-jobs | 16:50 |
| jeblair | pabelanger: and i think you mean 490977 | 16:50 |
| tobiash | jeblair: yay | 16:50 |
| pabelanger | 490977 yes | 16:51 |
| jeblair | tobiash: (i think you could also write "foo: null" with that change if you were so inclined) | 16:51 |
| tobiash | yes that should be the smae | 16:51 |
| tobiash | s/smae/same | 16:51 |
| pabelanger | jeblair: mordred: tobiash: do you mind adding https://review.openstack.org/#/c/490643/ to your review queue. Creates fetch-tox-output role that we can use from project-config | 17:00 |
| tobiash | pabelanger: lgtm | 17:03 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: DNM: test task with undefined variable https://review.openstack.org/490987 | 17:07 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: DNM: test task with undefined variable https://review.openstack.org/490987 | 17:09 |
| tobiash | there is a delay of 30s if a shell/command task fails with an ansible error like undefined variable: http://paste.openstack.org/show/617555/ | 17:12 |
| pabelanger | tobiash: not sure I understand | 17:18 |
| pabelanger | isn't that expected, since you added a sleep? | 17:18 |
| tobiash | pabelanger: not the sleep task (that was just to give me time to open the live log) | 17:18 |
| tobiash | pabelanger: when ansible bails out because of undefined variable the callback module waits on the streamer for 30s for a shell task that was never executed | 17:19 |
| pabelanger | [Zuul] Log Stream did not terminate | 17:19 |
| pabelanger | wonder if that is related some how | 17:20 |
| tobiash | jepp, that's related | 17:20 |
| tobiash | that is emitted after waiting for 30s for the streamer to end | 17:20 |
| pabelanger | IIRC, we still need to properly debug that | 17:20 |
| pabelanger | cool, didn't know that | 17:20 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: DNM: test task with exit 1 https://review.openstack.org/490987 | 17:23 |
| *** openstack has joined #zuul | 17:31 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat timer driver docs https://review.openstack.org/490951 | 17:31 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat sql driver docs https://review.openstack.org/490950 | 17:31 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat SMTP driver docs https://review.openstack.org/490947 | 17:31 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Use null values instead of the empty dict for sql reporter https://review.openstack.org/490978 | 17:31 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat tenant config docs https://review.openstack.org/490970 | 17:31 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat zuul driver docs https://review.openstack.org/490953 | 17:31 |
| jeblair | tobiash, pabelanger: i think mordred mentioned that may be happening on shell tasks that don't emit any output | 17:32 |
| tobiash | that's with a 'exit 1' task: http://paste.openstack.org/show/617557/ | 17:33 |
| tobiash | that returns immediatly | 17:33 |
| tobiash | so maybe this occurs if the shell was not started at all | 17:34 |
| tobiash | but I don't understand zuul_console yet | 17:34 |
| Shrews | Anyone know if it is possible to have sphinx docs link to a directory and list the files there? | 17:37 |
| tobiash | jeblair, pabelanger: I think I found the reason: http://git.openstack.org/cgit/openstack-infra/zuul/tree/zuul/ansible/library/zuul_console.py?h=feature/zuulv3#n185 | 17:44 |
| tobiash | if it fails to open the console file (and it will never exist) it gets into an endless loop trying to open it | 17:44 |
| tobiash | so we should probably add some timeout here (or is the retry needed?) | 17:45 |
| jeblair | tobiash: i'm not sure, but it may be possible for us to start trying to stream the log before ansible actually runs a command which would create it (even under normal conditions) | 17:50 |
| jeblair | which i think is why that's written that way | 17:50 |
| tobiash | jeblair: so retry limit 10 (5s)? | 17:50 |
| tobiash | or more? | 17:50 |
| jeblair | tobiash: i'd prefer to think of a more reliable way to deal with it | 17:51 |
| tobiash | jeblair: ok, different idea, can we distinguish between ansible parse error and shell-script error in zuul_stream? | 17:52 |
| tobiash | here: v2_runner_on_failed? | 17:52 |
| jeblair | tobiash: i don't know off the top of my head | 17:53 |
| tobiash | if yes, we could just kill the listener without timeout and leave the zuul_console as is | 17:53 |
| tobiash | jeblair: oh I think I was wrong, it seems to be on the reader side | 18:05 |
| tobiash | there is data from the stream | 18:06 |
| tobiash | but this is the only exit point of the reader: http://git.openstack.org/cgit/openstack-infra/zuul/tree/zuul/ansible/callback/zuul_stream.py?h=feature/zuulv3#n143 | 18:06 |
| tobiash | which requires the command to be executed and returned some exit code | 18:07 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul feature/zuulv3: Add project examples and doc references. https://review.openstack.org/491008 | 18:08 |
| Shrews | jeblair: i'm prepared to bikeshed on the proposed change there ^^^ as a potential way forward for our docs | 18:08 |
| Shrews | doesn't add much yet | 18:09 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: reformat metrics docs https://review.openstack.org/491010 | 18:11 |
| jeblair | Shrews: cool i'll take a look | 18:11 |
| Shrews | http://logs.openstack.org/08/491008/1/check/tox-docs/bc50199/html/user/config.html#job | 18:14 |
| pabelanger | Shrews: we could totally run a gate job using the examples too, if we wanted | 18:16 |
| pabelanger | to make sure they are always up to date | 18:16 |
| Shrews | that'd be kinda cool | 18:16 |
| Shrews | but, they'd probably have to remain openstack specific in that case. was trying to avoid that | 18:17 |
| jeblair | i'm kind of leaning the other way. there are certainly some nice things you get from having external examples, but i think it makes the documentation a little more difficult to write and maintain. for example it would be very easy for the highlighted lines to get out of sync. | 18:18 |
| jeblair | Shrews: would it be okay to just inline those examples? | 18:18 |
| jeblair | also, sorry i have to run real quick. biab. | 18:19 |
| Shrews | jeblair: we could, but then we'd have to keep the doc and source files in sync. not sure what doing that buys us | 18:19 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: DNM: test task with exit 1 https://review.openstack.org/490987 | 18:23 |
| *** adam_g has quit IRC | 18:24 | |
| *** adam_g has joined #zuul | 18:25 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul feature/zuulv3: Add project examples and doc references. https://review.openstack.org/491008 | 18:26 |
| *** openstack has joined #zuul | 18:30 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: DNM: test task with empty command https://review.openstack.org/490987 | 18:31 |
| jeblair | Shrews: back. what do you mean doc and source files in sync? i just meant put the contents of those example files directly into code-blocks in the rst files | 18:42 |
| Shrews | jeblair: My thought was that having a directory of different types of examples (I sort of inferred that you wanted something like this yesterday)? would be useful, even without the documentation. | 18:44 |
| jeblair | Shrews: oh, hrm. i only meant within the docs themselves. | 18:44 |
| Shrews | jeblair: I sort of like when projects include full examples that I can browse, independently of the docs | 18:44 |
| jeblair | Shrews: yes, that can be nice; though i'm not sure some of the same reasons apply here -- for instance, folks are very unlikely to be able to copy an example and start using it. | 18:46 |
| jeblair | whereas i think it will be very difficult to write and maintain the docs with all of the examples being external. | 18:48 |
| Shrews | jeblair: how so? | 18:50 |
| jeblair | Shrews: if a line is every added to one of those example files, the docs would need to be updated to change the line emphasis. but there's no indication that the examples directory is used by the docs. it seems quite likely that a change like that could slip through. | 18:51 |
| Shrews | we could mitigate some of that by moving the examples under the doc/ structure (i had it there originally, actually) | 18:52 |
| Shrews | but yeah, we'd have to be alert for highlight changes | 18:52 |
| jeblair | Shrews: moreover, any text that you write that refers to the examples means that you need to go look at probably several example files in order to make references (as opposed to just being able to look up a couple of paragraphs in the same file) | 18:53 |
| jeblair | Shrews: (put another way -- i *had* to render the docs in order to understand them, which is not something i'm used to. that extra effort reviewing is also extra effort writing.) | 18:54 |
| Shrews | ok, sounds like you're a strong "no" on this approach. i can just inline it | 18:54 |
| Shrews | no biggie :) | 18:54 |
| jeblair | Shrews: i wouldn't characterize it as a strong no, or at least not permanently so. more that i'm concerned we don't have the discipline for it at the moment. i think i'd like to revisit it later. | 18:57 |
| jeblair | Shrews: (and yeah, if we do revisit this, i think having it under the doc tree helps with that, and also with grepping) | 18:59 |
| jeblair | Shrews: regarding the content itself -- do you want to mention the project stanza under the job section? -- because we haven't gotten to 'project' yet in the narrative. what about putting the complete example under 'configuration items' then excerpting parts of it under the relevant items (so maybe job+playbook under job, then job+project under project)? | 19:07 |
| Shrews | jeblair: i'm not sure now. i want to rethink the approach, so just ignore what i have up right now. | 19:11 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: Try to early terminate streaming on ansible errors https://review.openstack.org/491027 | 19:12 |
| jeblair | Shrews: okiedokie | 19:12 |
| tobiash | jeblair, pabelanger: what do you think of ^^^ for fixing the 'stream did not terminate' problem? | 19:13 |
| jeblair | tobiash: i think that's a question for mordred too ^ | 19:13 |
| tobiash | oh, sure, sorry mordred | 19:14 |
| tobiash | that's the best I could come up after digging more than I wanted into the command ansible module | 19:14 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Support longer pkcs1-oaep secrets https://review.openstack.org/490620 | 20:05 |
| pabelanger | yay | 20:06 |
| *** jkilpatr_ has quit IRC | 20:41 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: change default title for value xref https://review.openstack.org/491091 | 20:55 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: enable show_related https://review.openstack.org/491097 | 21:15 |
| pabelanger | jeblair: just walked the stack of doc changes, looking really good! | 21:26 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Remove score attribute from sqlreporter config https://review.openstack.org/490972 | 21:29 |
| *** jkilpatr_ has joined #zuul | 21:29 | |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat SMTP driver docs https://review.openstack.org/490947 | 21:33 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: add very simple architecture diagram https://review.openstack.org/491100 | 21:46 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat sql driver docs https://review.openstack.org/490950 | 22:23 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: DNM - testing base-test https://review.openstack.org/491110 | 22:26 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: DNM - testing base-test https://review.openstack.org/491110 | 22:27 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Docs: add very simple architecture diagram https://review.openstack.org/491100 | 22:41 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-sphinx master: Add setup to module import https://review.openstack.org/491128 | 23:02 |
| mordred | tobiash, jeblair: yes - that approach looks good to me | 23:05 |
| mordred | jeblair: also, I agree with pabelanger - the docs stack is amazing | 23:06 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add zuul-sphinx as a requirement https://review.openstack.org/491129 | 23:07 |
| jeblair | pabelanger, mordred: oh cool, thanks :) | 23:07 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat timer driver docs https://review.openstack.org/490951 | 23:13 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat zuul driver docs https://review.openstack.org/490953 | 23:14 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat tenant config docs https://review.openstack.org/490970 | 23:14 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Use null values instead of the empty dict for sql reporter https://review.openstack.org/490978 | 23:15 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: reformat metrics docs https://review.openstack.org/491010 | 23:15 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Docs: change default title for value xref https://review.openstack.org/491091 | 23:17 |
| pabelanger | mordred: https://review.openstack.org/490643/ is ready for a review, created new fetch-tox-output role. We'll need that for our publish-openstack-python-tarball job | 23:21 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-sphinx master: Change usage from zuul_sphinx.zuul to zuul_sphinx https://review.openstack.org/491131 | 23:22 |
| openstackgerrit | Merged openstack-infra/zuul-sphinx master: Add setup to module import https://review.openstack.org/491128 | 23:22 |
| jeblair | 491131 is trivial; i'm going to self-approve it and make a release | 23:29 |
| mordred | jeblair: ++ | 23:30 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Set success-url for tox-docs job https://review.openstack.org/491132 | 23:34 |
| openstackgerrit | Merged openstack-infra/zuul-sphinx master: Change usage from zuul_sphinx.zuul to zuul_sphinx https://review.openstack.org/491131 | 23:34 |
| jeblair | pushed zuul-sphinx 0.1.3 | 23:35 |
| jeblair | pabelanger: nit on that, but should work | 23:36 |
| pabelanger | jeblair: ya, thanks. | 23:36 |
| pabelanger | jeblair: I am trying to decided if we want it setup in zuul-jobs or openstack-zuul-jobs. I'm leaning towards openstack-zuul-jobs, because we have openstack-docs-build running on zuul ATM | 23:37 |
| pabelanger | but, maybe zuul-jobs is the right place | 23:37 |
| pabelanger | since tox-docs is the parent | 23:37 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Set success-url for tox-docs job https://review.openstack.org/491132 | 23:38 |
| pabelanger | okay, sold my self on zuul-jobs | 23:38 |
| jeblair | pabelanger: yeah, it's an open question in my mind. :) it will affect us either way because it will be inherited. | 23:39 |
| pabelanger | ya, lets go with zuul-jobs for now. | 23:39 |
| clarkb | is the success url implicitly relative to the logs location? | 23:39 |
| clarkb | if so I think that is slight departure in behavior where we currently publish that stuff to docs-draft.openstack.org on a different "disk" than logs.openstack.org | 23:40 |
| jeblair | clarkb: i have good news for you! it's both! https://docs.openstack.org/infra/zuul/feature/zuulv3/user/config.html#attr-job.success-url | 23:40 |
| pabelanger | clarkb: ya, docs-draft.o.o is something we still need to do | 23:41 |
| jeblair | clarkb: it's relative if it's a relative url, or, make an absolute url and it takes over | 23:41 |
| clarkb | ah ok | 23:41 |
| clarkb | the biggest reason to have separate disks for that is variable retention policy though I think its 30 days for both right now :/ | 23:41 |
| jeblair | clarkb, pabelanger: however, it's likely that we'll probably just do something parallel for docs draft -- since in that case we can let the ansible publisher tell us where it put things. so i expect at least the base part of the docs-draft url to come from ansible just like the current stuff. | 23:42 |
| pabelanger | ya | 23:42 |
| jeblair | clarkb: maybe we should consider folding it back in. would get an extra TB of space. :| | 23:43 |
| jeblair | i'm guessing docs builds aren't as big as they used to be | 23:44 |
| clarkb | jeblair: worth considering. We can also likely consider bumping logs retention back up again | 23:44 |
| clarkb | we've got it under control now I think | 23:44 |
| clarkb | jeblair: because no more docbook? | 23:44 |
| jeblair | clarkb: that's my assumption, yes. | 23:45 |
| pabelanger | ya, logs.o.o looks to be at 9TB ATM | 23:45 |
| clarkb | fungi: what was your process of determining largest jobs? | 23:45 |
| clarkb | fungi: du -s on a change number prefix then sorting by job name? | 23:46 |
| jeblair | nah, i'm wrong. there hasn't really been a systemic change in doc-draft usage over the past year. | 23:46 |
| fungi | clarkb: great question! lemme see if i can dig up my notes but it was something along those lines | 23:46 |
| jeblair | just normal looking cycles. | 23:46 |
| clarkb | fungi: we might want to rerun that and tackle the next things on the list | 23:46 |
| clarkb | jeblair: dedup would probably help significantly on that fs | 23:46 |
| jeblair | clarkb: indeed | 23:46 |
| fungi | took somewhere between 24-48 hours to complete across the entire corpus, if memory serves | 23:47 |
| clarkb | we could live dangerously and use btrfs | 23:47 |
| clarkb | but red hat just announced its too dangerous for them so maybe not | 23:47 |
| clarkb | (which is slightly funny bceause xfs) | 23:48 |
| jeblair | please not the btrfs. i did that once it was not fun. | 23:48 |
| mordred | if jeblair doesn't behave we will subject him to the btrfs | 23:48 |
| pabelanger | maybe discuss at PTG about hashing my date for logs now? | 23:48 |
| pabelanger | I mean, we could start testing with zuulv3.o.o now | 23:48 |
| jeblair | mordred: not the comfyfs! | 23:48 |
| clarkb | I've been using it because it is suse default and its been fine. But I expect on older kernels its much sadness | 23:49 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Update the zuul-sphinx extension config https://review.openstack.org/491134 | 23:50 |
| clarkb | one thing that would be neat is fs scrubbing on merge | 23:50 |
| clarkb | its late on friday so we can talk about neat features that will never happen right? | 23:50 |
| clarkb | change merges, its doc drafts are all deleted | 23:50 |
| jeblair | clarkb: that's totally plausible in zuulv3 :) | 23:50 |
| clarkb | ya, I worry more about iops overload | 23:51 |
| jeblair | zero-node job in a change-merged pipeline with like 1 line of ansible :) | 23:51 |
| fungi | for i in `seq 20`;do du -sk /srv/static/logs/`echo $RANDOM|cut -c-2`/??????/*/*/* >> du_sample.txt;done | 23:51 |
| clarkb | (delete cron is laready a 2 day process iirc) | 23:51 |
| fungi | for j in `cut -d/ -f9 du_sample.txt|sort -u`;do echo $[$(grep /$j$ du_sample.txt|cut -f1|sed s/^/+/)] $j;done|sort -n|tail -n100>summary.txt;echo $[$(cut -f1 du_sample.txt |sed s/^/+/)] total kilobytes>>summary.txt | 23:51 |
| clarkb | jeblair: but we can give it a go and stop if the filesystem falls over | 23:51 |
| jeblair | clarkb: should give the delete cron much less work to do, so may be a wash as far as io | 23:52 |
| jeblair | basically, it's zero-sum io :) | 23:52 |
| clarkb | jeblair: ya distribute it over time too rather than all at once | 23:52 |
| jeblair | i feel like someone should give fungi some output. even if it's just a spinning cursor. | 23:53 |
| clarkb | thats quite the script | 23:53 |
| clarkb | fungi: if you haven't already done it I'll start the first command in a screen | 23:53 |
| fungi | clarkb: go for it. that'll take overnight probably | 23:54 |
| clarkb | fungi: does it need to run as root? | 23:54 |
| fungi | then once it's done (or really at any point along the way for that matter) you can run the other in another session to get preliminary analysis fairly quickly | 23:54 |
| fungi | yeah, i ran it under the root account | 23:54 |
| fungi | in part because it's easier for others to attach to a screen session started as root | 23:55 |
| clarkb | we serve logs under different account than we write them so they should be all readable but guessing there are corner cases where taht assumption is false | 23:55 |
| fungi | well, also root gets some priority scheduling, if my arcane knowledge of niceness levels is still applicable in this day and age | 23:55 |
| clarkb | ok thats running and I'm tailing the sample file in window 1 | 23:56 |
| fungi | attached | 23:57 |
| clarkb | fungi must hate attaching to screens I've started | 23:57 |
| clarkb | with my full screen terminals | 23:57 |
| fungi | if we used tmux it would just instantly resize its geometry to accommodate me ;) | 23:58 |
| fungi | that's one of the things i find impressive about tmux is that it scales columns and rows to whatever the least common denominator of attached sessions is | 23:59 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!