| *** evrardjp has quit IRC | 02:33 | |
| *** evrardjp has joined #zuul | 02:33 | |
| *** josefwells has quit IRC | 02:45 | |
| *** ykarel has joined #zuul | 04:44 | |
| *** bhavikdbavishi has joined #zuul | 04:54 | |
| *** bhavikdbavishi has quit IRC | 05:20 | |
| *** bhavikdbavishi has joined #zuul | 05:21 | |
| *** ykarel_ has joined #zuul | 05:36 | |
| *** ykarel has quit IRC | 05:38 | |
| *** bhavikdbavishi has quit IRC | 05:58 | |
| *** bhavikdbavishi has joined #zuul | 06:13 | |
| *** bhavikdbavishi1 has joined #zuul | 06:18 | |
| *** bhavikdbavishi has quit IRC | 06:20 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 06:20 | |
| *** bhavikdbavishi has quit IRC | 06:26 | |
| *** bhavikdbavishi has joined #zuul | 06:27 | |
| *** bhavikdbavishi has quit IRC | 06:39 | |
| *** bhavikdbavishi has joined #zuul | 07:03 | |
| *** ykarel_ is now known as ykarel | 07:12 | |
| *** jpena|off is now known as jpena | 07:30 | |
| *** bhavikdbavishi has quit IRC | 07:32 | |
| *** tosky has joined #zuul | 07:47 | |
| *** saneax has joined #zuul | 07:59 | |
| *** sshnaidm|afk is now known as sshnaidm|pto | 08:00 | |
| *** yoctozepto6 is now known as yoctozepto | 08:03 | |
| *** nils has joined #zuul | 08:22 | |
| *** ykarel is now known as ykarel|lunch | 08:25 | |
| *** mgoddard has quit IRC | 08:43 | |
| *** mgoddard has joined #zuul | 08:43 | |
| *** ykarel|lunch is now known as ykarel | 09:34 | |
| openstackgerrit | Lucas Alvares Gomes proposed zuul/zuul-jobs master: [dnm] testing devstack 791085 https://review.opendev.org/c/zuul/zuul-jobs/+/791117 | 09:37 |
|---|---|---|
| *** jpena is now known as jpena|lunch | 11:34 | |
| *** rlandy has joined #zuul | 11:49 | |
| *** bhavikdbavishi has joined #zuul | 12:14 | |
| *** rlandy is now known as rlandy|training | 12:14 | |
| *** bhavikdbavishi1 has joined #zuul | 12:19 | |
| *** bhavikdbavishi has quit IRC | 12:21 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 12:21 | |
| *** jpena|lunch is now known as jpena | 12:31 | |
| *** Shrews has joined #zuul | 12:36 | |
| *** saneax has quit IRC | 12:44 | |
| *** josefwells has joined #zuul | 12:51 | |
| *** rlandy|training is now known as rlandy | 13:23 | |
| *** bhavikdbavishi has quit IRC | 13:29 | |
| openstackgerrit | Lucas Alvares Gomes proposed zuul/zuul-jobs master: [dnm] testing devstack 791085 https://review.opendev.org/c/zuul/zuul-jobs/+/791117 | 13:39 |
| lyr | Hey, I tried to use zuul-client (0.0.4) enqueue-ref, error 500, on server side it's missing "trigger". Which was an existing option in "zuul" cli, and zuul-client seems to be missing it | 13:46 |
| lyr | I got zuul from software factory 3.5 | 13:48 |
| lyr | Maybe it's a change in recent zuul ? | 13:48 |
| *** Deknos has joined #zuul | 14:23 | |
| *** bhavikdbavishi has joined #zuul | 14:43 | |
| corvus | lyr: as of zuul 4.0, the rest api should no longer accept the trigger argument; are you using a new zuul-client with a zuul < 4.0? | 14:43 |
| lyr | I did pip install zuul-client | 14:43 |
| lyr | Zuul version: 3.19.1-3.el7 | 14:44 |
| corvus | lyr: i'm not sure zuul-client ever sent the trigger event, so in other words, i don't think 'zuul-client enqueue-ref' is ever worked before zuul 4.0 | 14:45 |
| *** bhavikdbavishi1 has joined #zuul | 14:45 | |
| corvus | s/trigger event/trigger parameter/ | 14:46 |
| *** bhavikdbavishi has quit IRC | 14:47 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 14:47 | |
| openstackgerrit | Lucas Alvares Gomes proposed zuul/zuul-jobs master: [dnm] testing devstack 791085 https://review.opendev.org/c/zuul/zuul-jobs/+/791117 | 15:08 |
| *** bhavikdbavishi has quit IRC | 15:15 | |
| corvus | zuul-maint: i pushed 4.3.0 | 15:28 |
| clarkb | cool so that gets us the git dir moves and some of the zk stuff but not the executor changes | 15:28 |
| corvus | yep | 15:29 |
| corvus | and i think the next batch of changes will be easy to roll back -- it's just gear -> zk | 15:30 |
| corvus | so if we mess it up again, a rollback to 430 should be no big deal | 15:30 |
| *** adpopescu has joined #zuul | 15:37 | |
| *** ykarel has quit IRC | 15:52 | |
| *** jamesmcarthur has joined #zuul | 15:56 | |
| openstackgerrit | Lucas Alvares Gomes proposed zuul/zuul-jobs master: [dnm] testing devstack 791436 https://review.opendev.org/c/zuul/zuul-jobs/+/791117 | 16:11 |
| *** rlandy is now known as rlandy|biab | 16:25 | |
| *** jpena is now known as jpena|off | 16:28 | |
| clarkb | corvus: I think the problem with the zuul executors is that they way you extend yaml's loader to recognize additional objects relies on side effects and to get those side effects you need to import zuul.configloader in any context wishing to have that additional yaml goodness | 16:42 |
| clarkb | corvus: we get that in our unittest because everything ends up imported, but when you run one of the more specialized daemons this isn't necessarily the case | 16:42 |
| clarkb | corvus: maybe that pkcs1 YAMLObject should be split into yamlutil and configloader can import it from there and others get it when using yamlutil | 16:42 |
| clarkb | (to be clear I mean the general 'you' here that seems to be the way pyyaml does it and there isn't a more explicit way that might make this more clear) | 16:46 |
| corvus | clarkb: oh nice, that sounds plausible. | 16:49 |
| corvus | clarkb: i think we *can* be more explicit, if we want to; but i think maybe for something like the pkcs thing having it be global is kind of nice. so maybe we just move that import around. | 16:50 |
| corvus | (er, move the object def into yamlutil as you suggest) | 16:51 |
| fungi | it does seem like a good explanation for why we ran into it where we did and not in testing | 16:51 |
| clarkb | corvus: ya | 16:51 |
| corvus | i'm going out for a bit; bbl | 16:51 |
| clarkb | corvus: also note that configloader.py has a couple of other override things that we may need to successfully parse configs elsewhere? | 16:52 |
| clarkb | (I am not sure how much of that gets pushed down to the executors) | 16:52 |
| corvus | clarkb: the others should be only local to configloader | 16:52 |
| corvus | they're about being able to do the nice error messages; shouldn't matter for the executor | 16:53 |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Move encrypted/pkcs1-oaep yaml object handling into yamlutil https://review.opendev.org/c/zuul/zuul/+/791514 | 16:59 |
| clarkb | That doesn't add any testing, but I figure a good first check is that the move works at all (it should but who knows). Maybe if the trivial move passes testing you can unrevert over the top of it and see what things look like | 17:00 |
| *** bhavikdbavishi has joined #zuul | 17:10 | |
| *** rlandy|biab is now known as rlandy | 17:17 | |
| *** jamesmcarthur has quit IRC | 17:34 | |
| mordred | clarkb: totally seems plausible to me too ... assuming you're right, how do we set up a test to make sure we don't break that in the future? | 17:42 |
| clarkb | mordred: I think maybe leaning on more functional testing that runs services independently of each other? Figuring that out in a unittest context is likely to be fragile and annoying | 17:43 |
| clarkb | just due to the way python works | 17:43 |
| mordred | yeah | 17:43 |
| mordred | I'm a little sad that the way the yaml extension works is a magic side-effect | 17:43 |
| clarkb | if we do want to go the more unittest like method wht we could do is write a few tests in a corner then run those separately from the main test run | 17:45 |
| *** jamesmcarthur has joined #zuul | 17:45 | |
| clarkb | that way we can more tightly control what is loaded into the test runner for those tests | 17:46 |
| clarkb | the tox.ini would end up looking like stestr run all the normal tests && stestr run the pyyaml tests | 17:46 |
| *** jamesmcarthur_ has joined #zuul | 17:47 | |
| *** jamesmcarthur has quit IRC | 17:51 | |
| *** smyers has quit IRC | 17:53 | |
| fungi | could the quickstart job test it effectively? | 17:54 |
| fungi | the services there are quite thoroughly isolated, and not too dissimilar from how we run things in the environment where we encountered the problem | 17:55 |
| clarkb | yup that may work if we pass jobs through it | 17:56 |
| josefwells | corvus: whats the status on the zuul-operator kopf stuff.. is it worth doing some testing on? | 17:59 |
| avass | josefwells: I think it's supposed to be ready | 18:02 |
| avass | but since the stack isn't merged they're still subject to change :) | 18:02 |
| josefwells | avass: sorry this is my unfamiliarity with gerrit, as it looks like it is on a branch "kopf", but I don't see that branch in the zuul-operator repo.. is there a way to grab that stack of proposed changes easily? | 18:04 |
| mordred | josefwells: the git review command has an option to download a change | 18:05 |
| mordred | so if you find the last change in the stack (https://review.opendev.org/c/zuul/zuul-operator/+/786349) | 18:05 |
| mordred | you can go into a zuul checkout and do "git review -d 786349" | 18:05 |
| mordred | and it will download it and put it in to a local branch for you | 18:06 |
| mordred | s/zuul/zuul-operator/ | 18:06 |
| avass | you can also just click the three dots on the top right and download the patch :) | 18:07 |
| mordred | yah - that too | 18:07 |
| josefwells | mordred: avass: Thanks, I'll see if I can get it going this way | 18:08 |
| clarkb | ok my change does not work and results in: "could not determine a constructor for the tag '!encrypted/pkcs1-oaep'" which is maybe good beacuse it is similar to the problem we saw on the executors | 18:21 |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Move encrypted/pkcs1-oaep yaml object handling into yamlutil https://review.opendev.org/c/zuul/zuul/+/791514 | 18:28 |
| clarkb | if ^ works better then we have to be very careful about aligning the SafeLoader implementation | 18:28 |
| mordred | isn't there a yamllib api that can be used to explicitly add things to a loader? | 18:33 |
| clarkb | mordred: there is https://pyyaml.org/wiki/PyYAMLDocumentation search Constructors, representers, resolvers | 18:47 |
| clarkb | thats probably a reasonable next step if we can't get this to look sane | 18:47 |
| mordred | yeah- yaml.add_constructor and yaml.add_representer is what I was thinking | 18:49 |
| *** jamesmcarthur_ has quit IRC | 18:59 | |
| *** Shrews has quit IRC | 19:00 | |
| *** rlandy is now known as rlandy|training | 19:03 | |
| *** smyers has joined #zuul | 19:09 | |
| clarkb | the py38 unittest job has passed on my second patchset. I think there may be two layers of issue here that this exposes. The first is the one I initially brought upwhere you need to import that object to side effect its registration with pyyaml. The other is one of the object attributes is the SafeLoader implementation to apply it against and you need to ensure that you apply it to the load that you | 19:39 |
| clarkb | are going to use everywhere | 19:39 |
| *** bhavikdbavishi has quit IRC | 19:41 | |
| *** rlandy|training is now known as rlandy | 20:16 | |
| *** ikhan has joined #zuul | 20:28 | |
| *** ikhan has quit IRC | 20:47 | |
| *** adpopescu has quit IRC | 20:49 | |
| *** ikhan has joined #zuul | 21:00 | |
| *** josefwells has quit IRC | 21:12 | |
| *** ikhan has quit IRC | 21:59 | |
| *** ikhan has joined #zuul | 22:03 | |
| *** rlandy has quit IRC | 22:21 | |
| *** ikhan has quit IRC | 22:21 | |
| *** dpawlik has quit IRC | 22:45 | |
| corvus | clarkb, mordred: it may be worth switching to the more explicit method. | 22:47 |
| corvus | i have some code for that laying around | 22:50 |
| *** dpawlik7 has joined #zuul | 22:53 | |
| *** tosky has quit IRC | 23:11 | |
| *** sduthil has quit IRC | 23:28 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!