| *** saneax has quit IRC | 00:13 | |
| *** rfolco has joined #zuul | 00:15 | |
| *** swest has quit IRC | 01:14 | |
| *** swest has joined #zuul | 01:28 | |
| *** jamesmcarthur has joined #zuul | 01:40 | |
| *** jamesmcarthur has quit IRC | 02:10 | |
| *** irclogbot_1 has quit IRC | 02:17 | |
| *** altlogbot_2 has quit IRC | 02:19 | |
| *** irclogbot_3 has joined #zuul | 02:21 | |
| *** altlogbot_0 has joined #zuul | 02:22 | |
| *** openstackgerrit has joined #zuul | 02:36 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul master: web: add version to info endpoint https://review.opendev.org/609571 | 02:36 |
|---|---|---|
| *** bhavikdbavishi has joined #zuul | 03:04 | |
| *** bjackman has quit IRC | 03:32 | |
| *** bjackman has joined #zuul | 04:04 | |
| *** badboy has joined #zuul | 04:07 | |
| *** bhavikdbavishi1 has joined #zuul | 04:15 | |
| *** bhavikdbavishi has quit IRC | 04:16 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 04:16 | |
| *** bhavikdbavishi has quit IRC | 04:42 | |
| *** bhavikdbavishi has joined #zuul | 04:43 | |
| *** pcaruana has joined #zuul | 05:03 | |
| *** saneax has joined #zuul | 05:26 | |
| *** saneax has quit IRC | 05:31 | |
| *** bhavikdbavishi1 has joined #zuul | 05:47 | |
| *** bhavikdbavishi has quit IRC | 05:48 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 05:48 | |
| *** lennyb has joined #zuul | 06:30 | |
| *** raukadah is now known as chandankumar | 06:46 | |
| tristanC | Shrews: hello, do you know if this can be fixed with the current static driver design, or should we tell users to create a provider per label? https://review.opendev.org/662954 | 06:53 |
| *** saneax has joined #zuul | 07:07 | |
| *** bhavikdbavishi has quit IRC | 07:28 | |
| *** jangutter has joined #zuul | 07:29 | |
| *** jpena|off is now known as jpena | 07:30 | |
| openstackgerrit | Merged zuul/nodepool master: Explicitly set use_direct_get to False https://review.opendev.org/663368 | 07:43 |
| *** bhavikdbavishi has joined #zuul | 07:50 | |
| *** zbr has joined #zuul | 07:53 | |
| *** flepied has joined #zuul | 07:54 | |
| *** zbr is now known as zbr|ruck | 07:55 | |
| *** bhavikdbavishi has quit IRC | 08:03 | |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Reduce gear logging in tests https://review.opendev.org/664219 | 08:05 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Safely add Ansible password lookup plugin https://review.opendev.org/662870 | 09:04 |
| tobiash | fungi: that should fix the test ^ | 09:04 |
| openstackgerrit | Sorin Sbarnea proposed zuul/zuul-jobs master: Adds tox-mol job https://review.opendev.org/664230 | 09:06 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul master: cli: prevent exception for the show command https://review.opendev.org/664235 | 09:40 |
| openstackgerrit | Sorin Sbarnea proposed zuul/zuul-jobs master: Adds tox-mol job https://review.opendev.org/664230 | 09:54 |
| *** bhavikdbavishi has joined #zuul | 09:58 | |
| *** sshnaidm|off is now known as sshnaidm | 10:10 | |
| zbr|ruck | fungi: tobiash if you can review ^ it would be great. that is based on recommendation to add molecule job definition in zuul-jobs instead of openstack-zuul-jobs because is not really specific to openstack, could apply to any project. | 10:18 |
| openstackgerrit | Matthieu Huin proposed zuul/zuul master: [WIP] admin REST API: zuul-web integration https://review.opendev.org/643536 | 10:53 |
| *** sanjayu_ has joined #zuul | 11:03 | |
| *** saneax has quit IRC | 11:06 | |
| *** jpena is now known as jpena|lunch | 11:29 | |
| *** bhavikdbavishi has quit IRC | 11:38 | |
| *** bhavikdbavishi has joined #zuul | 11:39 | |
| *** badboy has quit IRC | 12:16 | |
| *** sanjayu_ has quit IRC | 12:40 | |
| Shrews | tristanC: i don't think so. that's just the way the algorithm works. you can certainly do 1 pool per label though | 12:58 |
| *** jpena|lunch is now known as jpena | 13:02 | |
| *** bjackman has quit IRC | 13:04 | |
| mordred | zbr|ruck: that looks great to me - maybe we should call it tox-molecule though? | 13:06 |
| *** bhavikdbavishi has quit IRC | 13:08 | |
| *** saneax has joined #zuul | 13:20 | |
| *** michael-beaver has joined #zuul | 13:34 | |
| fungi | tobiash: oh, thanks!!! | 13:35 |
| *** bjackman has joined #zuul | 13:46 | |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Safely add Ansible password lookup plugin https://review.opendev.org/662870 | 14:05 |
| zbr|ruck | mordred: i went for "mol" because is quick to type and happens to be the international abreviation for molecule. I just wanted to make very easy for users to run it often. nobody complied so far. | 14:19 |
| mordred | nod | 14:19 |
| zbr|ruck | mordred: i know what you mean. i can change it..., but lets see what others think. | 14:20 |
| mordred | yah - I agree, more input is a good thing there | 14:20 |
| mordred | thanks for writing that job, I think it's a good idea of a thing to have | 14:21 |
| zbr|ruck | mordred: i already see abreviations being popular -py37, -el7 -f28 also on jobs. | 14:21 |
| zbr|ruck | if it would up to me I would even rename linters to lint for the same reasons ;) | 14:23 |
| fungi | the only use of "mol" i know about is the abbreviation for molar mass: https://en.wikipedia.org/wiki/Molar_mass | 14:24 |
| fungi | granted, been a long time since i sat in a chem class | 14:24 |
| clarkb | these don't tend to be something people have to type a lot and if we learned anything from "dsvm" its that clarity in job name is a good thing | 14:28 |
| *** saneax has quit IRC | 14:30 | |
| openstackgerrit | David Shrewsbury proposed zuul/zuul master: Auto-delete expired autohold requests https://review.opendev.org/663762 | 14:33 |
| openstackgerrit | David Shrewsbury proposed zuul/zuul master: WIP: Mark nodes as USED when deleting autohold https://review.opendev.org/664060 | 14:33 |
| corvus | tobiash: minor comment on https://review.opendev.org/662818 -- is that worth an update? | 14:40 |
| *** jamesmcarthur has joined #zuul | 14:48 | |
| pabelanger | TIL: another ansible based 'CI/CD system' (eg: ansible as a service): https://playbook.cloud/ seen some ansible humans talking about it | 14:49 |
| clarkb | Shrews: ^ held nodes have already had their jobs complete on them so there is nothing to transition the nodes from used to deleting except for the node timeout (8 hours by default iirc) | 14:49 |
| pabelanger | zuul is also listed there | 14:49 |
| clarkb | Shrews: should that transition straight to delete instead? | 14:49 |
| clarkb | I guess delete doesn'twork in the case of static nodes | 14:50 |
| Shrews | clarkb: nodepool cleans up held nodes periodically, as long as they are not locked | 14:52 |
| *** bjackman has quit IRC | 14:52 | |
| corvus | we don't currently have a 'delete' state anymore | 14:52 |
| corvus | so yeah, i think used is most appropriate now | 14:53 |
| clarkb | corvus: there is still a state that tells the cleanup thread it can delete the nodes right? and that transition hasppens from used -> something when a job ends and the nodes aren't held | 14:53 |
| Shrews | clarkb: oh, i mean, USED not HELD | 14:53 |
| clarkb | Shrews: ya its on an 8 hour timeout by default iirc | 14:53 |
| corvus | clarkb: i believe that state is "used && unlocked" | 14:53 |
| clarkb | ah | 14:53 |
| Shrews | clarkb: held nodes may or may not have an expiration. i don't think we set a default in our config | 14:54 |
| Shrews | which i thik means they linger forever by default | 14:54 |
| Shrews | think | 14:54 |
| clarkb | if in a held state they do. The 8 hour timeout is for straggler nodes that haven't otherwise been deleted (I think it affects ready and used nodes) | 14:55 |
| clarkb | though maybe that isn't a thing anymore with the better db state via locks | 14:55 |
| Shrews | https://zuul-ci.org/docs/nodepool/configuration.html#attr-max-hold-age | 14:56 |
| Shrews | clarkb: is that what you're thinking of? ^^ | 14:56 |
| Shrews | or are we thinking of different things? | 14:57 |
| corvus | hold with no expiration means "keep forever". hold with expiration means "delete after expiration". there are no locks for hold. | 14:57 |
| corvus | let me revise that | 14:57 |
| corvus | hold with no expiration on the node means "use the default nodepool expiration time", if the default nodepool expiration time is unset, it means "keep forever". | 14:58 |
| clarkb | Shrews: no we have/had a thing that would delete nodes > 8 hours old if they are not currently held or being used by a job | 14:58 |
| Shrews | i think that's max-ready-age | 14:58 |
| clarkb | Shrews: this would cleanup ready nodes (to ensure we had up to date ready nodes on current images) as well as retrying deletes on things that hadn't deleted yet | 14:58 |
| Shrews | https://zuul-ci.org/docs/nodepool/configuration.html#attr-labels.max-ready-age | 14:59 |
| clarkb | but I'm not finding it anymore so we may have made it more specific to ready nodes now | 14:59 |
| Shrews | that doesn't have anything to do with retrying deletes though | 14:59 |
| clarkb | ya Ithink this code changed somewhere along the line | 14:59 |
| clarkb | it used to be a global check in a cleanup thread | 14:59 |
| clarkb | Is this node > 8 hours old? is it held or being used by a job? if not then delete it | 14:59 |
| zbr|ruck | mordred: what do you think about last comment on https://review.opendev.org/#/c/664230/2/zuul.yaml@219 ? (urls) | 15:01 |
| corvus | tobiash: +2 with comment on https://review.opendev.org/660878 | 15:07 |
| *** bjackman has joined #zuul | 15:13 | |
| *** jamesmcarthur has quit IRC | 15:16 | |
| *** bjackman_ has joined #zuul | 15:20 | |
| *** bjackman has quit IRC | 15:22 | |
| tobiash | corvus: re 662818, you're right, I'll update this | 15:24 |
| *** jamesmcarthur has joined #zuul | 15:29 | |
| *** bjackman_ has quit IRC | 15:37 | |
| openstackgerrit | Merged zuul/nodepool master: openstack: document the key-name parameter https://review.opendev.org/661677 | 15:44 |
| openstackgerrit | Merged zuul/zuul master: Use nodejs v10 in testing https://review.opendev.org/662339 | 15:57 |
| *** mattw4 has joined #zuul | 16:07 | |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Remove extra task from download-artifacts https://review.opendev.org/664391 | 16:10 |
| openstackgerrit | Merged zuul/zuul master: Update axios version and yarn.lock https://review.opendev.org/662316 | 16:10 |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Remove extra task from download-artifacts https://review.opendev.org/664391 | 16:11 |
| corvus | ofosos: it looks like the new tests in https://review.opendev.org/662134 are on the right track. let me know if you have any questions or need another set of eyes on that. meanwhile, i'll probably wait until that's ready before re-reviewing the stack (i don't anticipate major changes based on my previous reviews) | 16:20 |
| ofosos | corvus: stuff is still not working, but the problem is that I Don't get the same output locally as I do on the build farm | 16:22 |
| ofosos | Today is a public holiday so I'll try to debug that tomorrow | 16:23 |
| ofosos | Thanks for the kind words :) | 16:23 |
| openstackgerrit | Sorin Sbarnea proposed zuul/zuul-jobs master: Adds tox-mol job https://review.opendev.org/664230 | 16:25 |
| *** igordc has joined #zuul | 16:32 | |
| tobiash | corvus: responded on 663015 | 16:42 |
| corvus | tobiash: yeah, what i was thinking though is that with ansible storing stuff in ansible_tmp_dir, if it gets unmounted it may be hard to debug | 16:43 |
| tobiash | corvus: we could change that, but I never saw ansible leaving stuff behind yet though | 16:44 |
| corvus | tobiash: oh, so you don't think there's anything of value there? then okay, patch lgtm as-is. :) | 16:44 |
| tobiash | yes, ansible is supposed to put and delete its per task tempfiles there | 16:45 |
| corvus | yeah, i was thinking things like facts, but we use a different directory for the fact cache... | 16:45 |
| corvus | +2 | 16:46 |
| tobiash | :) | 16:46 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Parallelize github event processing https://review.opendev.org/662818 | 16:48 |
| tobiash | corvus: your comment on 660878 is meant for deployments that offloaded the js stuff from zuul-web? | 16:53 |
| *** jpena is now known as jpena|off | 16:53 | |
| openstackgerrit | Merged zuul/zuul-jobs master: Remove extra task from download-artifacts https://review.opendev.org/664391 | 16:54 |
| corvus | tobiash: yeah, anytime they're out of sync (i suppose even a from-source deployment which did not offload might be able to serve updated static files?) at any rate, you can view the results in the test job | 16:54 |
| tobiash | corvus: I can talk to the author ans ask him to switch that | 16:54 |
| corvus | it will certainly be the case for opendev | 16:54 |
| tobiash | otherwise this will probably break the ui for opendev until zuul-web has been restarted | 16:55 |
| corvus | it doesn't really bother me that much... i'd be okay merging it as-is... i mostly brought it up in case someone feels more strongly about it | 16:55 |
| corvus | tobiash: "break" is a strong word... it will just be inaccurate :) | 16:55 |
| tobiash | yes ;) | 16:55 |
| corvus | it won't actually crash with an error or anything | 16:55 |
| openstackgerrit | Merged zuul/zuul master: Annotate builds with event id https://review.opendev.org/658895 | 17:00 |
| tobiash | corvus: is 662316 already completely rolled out in opendev? | 17:02 |
| tobiash | looks like live log streaming is broken for me | 17:02 |
| tobiash | jepp, that breaks it :( | 17:03 |
| corvus | oops, i guess we should have checked that in the preview build | 17:04 |
| corvus | clarkb: ^ should we revert that? | 17:05 |
| tobiash | xterm.js is broken with that | 17:07 |
| tobiash | maybe there is an update for that | 17:07 |
| tobiash | hrm, it already took the newest | 17:08 |
| clarkb | Ya we should probably revert | 17:08 |
| corvus | i'll propose a revert | 17:08 |
| openstackgerrit | James E. Blair proposed zuul/zuul master: Revert "Update axios version and yarn.lock" https://review.opendev.org/664399 | 17:09 |
| tobiash | clarkb, corvus: a cap of xterm will do it for now, I'll prepare a change in a bit | 17:18 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Cap xterm.js to < 3.14 https://review.opendev.org/664404 | 17:22 |
| tobiash | clarkb, corvus: I tried this locally and it fixes xterm.js ^ | 17:22 |
| tobiash | so I think we won't need to revert | 17:23 |
| fungi | either solution seems good to me | 17:25 |
| tobiash | this seems to break it in xterm.js: https://github.com/xtermjs/xterm.js/pull/2094 | 17:26 |
| tobiash | went into 3.14 | 17:26 |
| openstackgerrit | Tobias Henkel proposed zuul/zuul master: Update cached repo during job startup only if needed https://review.opendev.org/648229 | 17:46 |
| tjgresha | does anyone have a link to docs on nodepool logging config? there is a reference to setting the path to the file here - https://zuul-ci.org/docs/nodepool/operation.html - but what goes in the file? | 18:02 |
| clarkb | tjgresha: it is a python logging config file. https://docs.python.org/3.7/library/logging.config.html#configuration-file-format I think we are trying to get away from people using that though and rely on sane defaults | 18:08 |
| clarkb | tobiash: corvus fungi I've approved the xterm.js cap | 18:10 |
| clarkb | if we want to go with the revert we can WIP the xterm cap and approve the other change | 18:10 |
| fungi | at least the cap lets us roll everything else forward (here's hoping there are no other regressions in those deps) | 18:11 |
| corvus | that's fine -- i was going to wait until check jobs came back so we could test the fix. but i believe tobiash that it works. it's probably fine. :) | 18:12 |
| clarkb | oh I can remove my +A if you want to double check | 18:12 |
| tjgresha | clarkb: i need insane logging as my my nodepool is still tearing down nodes as soon as they are created with a min ready set - i need a bigger hint as to what np doesn't like with the openstack cloud it is deploying to | 18:13 |
| clarkb | tjgresha: https://opendev.org/opendev/puppet-nodepool/src/branch/master/templates/nodepool-launcher.logging.conf.erb is what opendev uses iirc | 18:14 |
| openstackgerrit | David Shrewsbury proposed zuul/zuul master: Mark nodes as USED when deleting autohold https://review.opendev.org/664060 | 18:14 |
| clarkb | tjgresha: but you should be able to set debug logging on and get a ton of logs | 18:14 |
| clarkb | tjgresha: does setting the debug flag not produce enough data? | 18:14 |
| openstackgerrit | Clark Boylan proposed zuul/nodepool master: Exclude SDK with broken image upload tasks https://review.opendev.org/664415 | 18:17 |
| clarkb | corvus: Shrews mordred ^ fyi | 18:17 |
| mordred | clarkb: nod. we'll get to the bottom of that | 18:17 |
| Shrews | corvus: mordred: if 664060 passes, i think that's the end of the autohold revamp change. There's still some possible improvement around organizing the held nodes in the zk record, but we can improve on that later. | 18:18 |
| mordred | Shrews: sweet | 18:18 |
| clarkb | mordred: the story linkd there has what I think may be the issue | 18:18 |
| clarkb | mordred: the content type isn't being set properly | 18:18 |
| Shrews | clarkb: so 0.27 worked? | 18:18 |
| clarkb | Shrews: yes | 18:18 |
| Shrews | cool | 18:18 |
| clarkb | so I've downgraded nb02 now too | 18:18 |
| clarkb | mordred: I was going to push a patch upas I think this may be a small fix | 18:19 |
| mordred | clarkb: oh - cool | 18:19 |
| clarkb | the code that sets the content type was refactored too, it was just not called so I was going to update that and then someone that knows how to test this better can help sanity check it | 18:19 |
| mordred | ++ - I'll make sure to point gtema at it when next he's online | 18:20 |
| tjgresha | clarkb: you mean nodepool-launcher with the --debug ? | 18:21 |
| clarkb | tjgresha: ya | 18:21 |
| Shrews | tjgresha: hrm, could be a few things happening, but most likely nodepool isn't able to ssh into your nodes to verify them. debug logs should help | 18:23 |
| fungi | also rapid churn can sometimes correspond to a configuration where nodepool is unable to identify a "public" address for the node | 18:24 |
| fungi | for example if the cloud isn't calling the interfaces something which the sdk is able to guess as the public one | 18:24 |
| tjgresha | shrews: its deleting pretty quick and i am setting launch and boot times outs really high as i thought maybe that was the issue.. assume it would try to ssh a couple times before giving up .. | 18:28 |
| Shrews | i think it tries 3 times, iirc. could be wrong | 18:28 |
| * mordred is guessing that it's not getting to ssh and instead is showing up without an interface_ip | 18:29 | |
| Shrews | but if it can't even establish a connection, probably the ip thing | 18:29 |
| mordred | yeah | 18:29 |
| fungi | mordred: yeah, that's also my guess. we've seen at least a couple deployments in the past week where the people standing them up had that problem | 18:29 |
| fungi | having to do with choice of name for the network or inability to identify a need to create a floating ip in one situation where the cloud was deployed to require nat | 18:30 |
| clarkb | mordred: I'm not seeing an obvious way to set a header on a specific request because headers are apparnetly resource object attributes so they will apply to all requests of a resource? | 18:38 |
| clarkb | How does that even work with swift | 18:38 |
| mordred | clarkb: swift is different | 18:39 |
| clarkb | well swift and glance are alike in this particular regard seems like | 18:39 |
| tjgresha | thanks for help folks | 18:40 |
| mordred | clarkb: where's the request you're looking at? | 18:40 |
| clarkb | mordred: old: https://opendev.org/openstack/openstacksdk/src/tag/0.27.0/openstack/image/v2/_proxy.py#L277-L280 current: https://opendev.org/openstack/openstacksdk/src/branch/master/openstack/image/v2/_proxy.py#L277-L313 | 18:41 |
| clarkb | I thought the fix was going to be as simple as switch out self.update_image() with self._update_image_properties and add the header to the request | 18:43 |
| clarkb | but I don't think I can do that without affecting requests to subsequent operations on that resource (which we don't want) | 18:43 |
| clarkb | (at least I don't think we want that) | 18:43 |
| mordred | hrm. I think we might awnt to override commit for the image resource | 18:44 |
| *** jamesmcarthur has quit IRC | 18:44 | |
| mordred | ew though - that's gross | 18:44 |
| mordred | hrm. I wonder ... | 18:45 |
| mordred | nope. yeah - this is going to require a little more plumbing | 18:46 |
| clarkb | http://logs.openstack.org/04/664404/1/check/zuul-build-dashboard/da1a37a/npm/html/status | 18:55 |
| clarkb | seems to stream logs | 18:55 |
| fungi | tobiash: based on the timeout at http://logs.openstack.org/70/662870/5/check/tox-py36/0b21ec2/job-output.txt.gz#_2019-06-10_14_36_09_679858 do you think we need to up the timeout value at https://review.opendev.org/#/c/662870/5/tests/unit/test_v3.py line 2688? | 19:10 |
| tobiash | fungi: that test is a very long running test so it's plausible that it just timeouted on a busy node. I think increasing it makes sense | 19:13 |
| fungi | thanks for confirming. i'm increasing the number of jobs in it by almost 50% so increasing the timeout in that change is reasonable i guess. i'll try to resolve that and add the bit to create the file for the read test inside the work root so we don't need the ansible version checks | 19:15 |
| openstackgerrit | Merged zuul/zuul master: Cap xterm.js to < 3.14 https://review.opendev.org/664404 | 19:18 |
| fungi | is there an ansibleish way to write a string out to a file, or do i need to resort to copying or templating it from some other file shipped alongside the playbook? (or should i just use a shell/command task for that?) | 19:31 |
| fungi | the file module only seems to allow for creating empty files, and has no mechanism for injecting content into them that i can see | 19:33 |
| corvus | fungi: 'copy' | 19:33 |
| Shrews | or https://docs.ansible.com/ansible/latest/modules/lineinfile_module.html#lineinfile-module | 19:33 |
| fungi | ooh, thanks corvus! indeed, the copy module seems to have a content parameter | 19:34 |
| fungi | that'll do nicely | 19:34 |
| openstackgerrit | Jeremy Stanley proposed zuul/zuul master: Safely add Ansible password lookup plugin https://review.opendev.org/662870 | 19:37 |
| fungi | Shrews: thanks, linefile looks more useful for modifying existing files, but copy seems like a more straightforward way to just create a file with some arbitrary content | 19:38 |
| fungi | i didn't even think to look at it, because it didn't dawn on me it might let you copy from a strnig | 19:38 |
| Shrews | fungi: lineinfile has a create option, but whatever works | 19:38 |
| fungi | yeah, linefile in create mode is almost as simple | 19:39 |
| corvus | fungi: sorry for the short reply; there was a sandwich involved | 19:41 |
| fungi | short reply and a long sandwich! ;) | 19:41 |
| fungi | my replies are inversely proportional to the length of my sandwiches too | 19:42 |
| corvus | physical laws of sandwiches | 19:43 |
| Shrews | i want corvus' sandwich. is there a transitive property? | 19:43 |
| openstackgerrit | Merged zuul/nodepool master: Exclude SDK with broken image upload tasks https://review.opendev.org/664415 | 19:43 |
| corvus | Shrews: you don't want it anymore | 19:43 |
| fungi | ahh, negation property | 19:52 |
| openstackgerrit | Merged zuul/zuul master: Log github requests with annotated events https://review.opendev.org/660800 | 19:59 |
| Shrews | grrr, found a logic problem with the autohold changes. -1'd one of the changes | 20:00 |
| openstackgerrit | Merged zuul/zuul master: Annotate logs around build completion and cancellation https://review.opendev.org/660806 | 20:07 |
| mordred | clarkb: so ... there's code to handle the content type and it looks like it's plumbed through ... so I'm still digging | 20:24 |
| *** jamesmcarthur has joined #zuul | 20:26 | |
| fungi | reminder to self: don't try to pull up the run task result from zuul unit tests in ara | 20:29 |
| *** pcaruana has quit IRC | 20:31 | |
| openstackgerrit | Jeremy Stanley proposed zuul/zuul master: Safely add Ansible password lookup plugin https://review.opendev.org/662870 | 20:35 |
| mordred | clarkb: are you sure that that header wasn't being sent? I'm going to try a manual upload from nb01 from a virtualenv so I can get a full trace | 20:38 |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Correct artifact return url for python/js https://review.opendev.org/664437 | 20:42 |
| clarkb | mordred: no I just noticed that that was like the only thing that changed. The rest of the metadata stuff seemed to be mostly the same | 20:42 |
| clarkb | mordred: possible that the call path isn't hitting the bit to update the header? | 20:42 |
| mordred | clarkb: I can't see any way for that to happen, unfortunately | 20:43 |
| clarkb | mordred: the accept header update there is new | 20:44 |
| clarkb | compared to https://opendev.org/openstack/openstacksdk/src/tag/0.27.0/openstack/image/v2/_proxy.py#L277-L278 | 20:44 |
| mordred | clarkb: maybe it's the add of the accept header | 20:45 |
| *** jamesmcarthur has quit IRC | 20:55 | |
| *** jamesmcarthur has joined #zuul | 20:58 | |
| *** mattw4 has quit IRC | 21:04 | |
| *** mattw4 has joined #zuul | 21:04 | |
| openstackgerrit | Merged zuul/zuul-jobs master: Correct artifact return url for python/js https://review.opendev.org/664437 | 21:07 |
| *** jamesmcarthur has quit IRC | 21:08 | |
| *** jamesmcarthur has joined #zuul | 21:20 | |
| *** mattw4 has quit IRC | 21:31 | |
| *** mattw4 has joined #zuul | 21:32 | |
| fungi | topic:password-lookup seems to be ready to go now if anyone has time. it fixes a (probably low-risk) known security design flaw in the run-buildset-registry role | 21:54 |
| fungi | the zuul-jobs change isn't going to pass its zuul-jobs-test-registry job until the zuul change it depends on is deployed on opendev's executors, but i'll recheck it once that's the case | 21:55 |
| corvus | fungi: can you call the parse_parameters method directly? | 21:57 |
| *** panda has quit IRC | 21:58 | |
| corvus | i know some of the plugins do call ansible helper methods for similar sorts of things | 21:58 |
| fungi | it's not actually a method, if you look at the source | 21:59 |
| fungi | it's a function instantiated locally in the same module which also houses the class | 21:59 |
| fungi | we probably could call if it we import the module | 22:00 |
| fungi | happy to entertain alternatives. it was kinda ugly to recreate it, i know | 22:00 |
| *** panda has joined #zuul | 22:01 | |
| corvus | fungi: that seems to err on the side of being more restrictive while being simple, so probably ok. though i think calling "password._parse_parameters(term)" would work, cause i think the original plugin is imported as 'password' | 22:03 |
| fungi | oh, indeed, it's password.LookupModule we're using as the module class | 22:04 |
| fungi | i'll give that a shot | 22:04 |
| corvus | fungi: yeah, at some point we worked out all the magic needed to do that import and stuck it in that function | 22:04 |
| corvus | it was not easy, but seems to have held up so far | 22:05 |
| openstackgerrit | Jeremy Stanley proposed zuul/zuul master: Safely add Ansible password lookup plugin https://review.opendev.org/662870 | 22:09 |
| fungi | corvus: that ^ ought to do it | 22:09 |
| fungi | we could stand to do an audit of the other allowed lookup plugins. not all of them are being tested... though testing them this way may not scale given i had to bump the timeout by 90 seconds just to add reasonably thorough scenarios for one lookup plugin | 22:11 |
| corvus | yeah. we could shard the test (we could probably very easily shard it 1:1, but it's an expensive test involving a full ansible invocation, so that may increase the overall runtime quite a bit). | 22:15 |
| *** jamesmcarthur has quit IRC | 22:15 | |
| fungi | i haven't looked closely at any other than the ini plugin you pointed me to as an example, and as we discovered its methods we're shadowing were removed after ansible 2.3 | 22:19 |
| *** jamesmcarthur has joined #zuul | 22:20 | |
| mordred | clarkb: http://paste.openstack.org/raw/752708/ - there's the request along with the response | 22:25 |
| mordred | we're sending the content-type correctly - and the body does look like it's a proper json patch | 22:25 |
| *** jamesmcarthur has quit IRC | 22:25 | |
| clarkb | mordred: how hard is it to get that for a 0.27.0 request then we can compare | 22:27 |
| mordred | not SUPER hard - just long (you have to actually upload a valid image to get to this point in the process) | 22:27 |
| * clarkb reads about empty accept header | 22:27 | |
| mordred | clarkb: I've got a run going right now patched to not pass that empty accept header - just on a hunch | 22:27 |
| mordred | if that doesn't fix it, I'll do it again with 0.27.0 | 22:28 |
| clarkb | seems that its kind of a grey area | 22:29 |
| clarkb | but I could totally see jetty being mad about it | 22:29 |
| *** jamesmcarthur has joined #zuul | 22:29 | |
| mordred | yah | 22:29 |
| clarkb | (also wow jetty) | 22:32 |
| mordred | right? | 22:33 |
| mordred | clarkb: there is a stackoverflow comment from a person who seems to think 400 would be an acceptable reponse if one reads the spec in that way | 22:33 |
| openstackgerrit | Kevin Carter (cloudnull) proposed zuul/zuul-jobs master: Add multi-distro support to install-docker https://review.opendev.org/664455 | 22:39 |
| SpamapS | I love how RFC's and stackoverflow work similarly to the US constitution and Supreme Court | 22:40 |
| fungi | the pirates code is really more of a guideline | 22:41 |
| mordred | clarkb: accept removal did not work | 22:54 |
| mordred | clarkb: I'm going to try 0.27.0 now | 22:54 |
| mordred | I'll let you know how it goes in about 27 minutes | 22:55 |
| *** jamesmcarthur has quit IRC | 22:59 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed zuul/zuul-jobs master: Add upstream docker installation https://review.opendev.org/664463 | 23:18 |
| openstackgerrit | Kevin Carter (cloudnull) proposed zuul/zuul-jobs master: Add multi-distro support to install-docker https://review.opendev.org/664455 | 23:23 |
| clarkb | mordred: any luck? | 23:26 |
| openstackgerrit | Kevin Carter (cloudnull) proposed zuul/zuul-jobs master: Add multi-distro support to install-docker https://review.opendev.org/664455 | 23:37 |
| *** threestrands has joined #zuul | 23:44 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed zuul/zuul-jobs master: Add multi-distro support to install-docker https://review.opendev.org/664455 | 23:46 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!