| airship-irc-bot | <raliev> well, we are still experiencing TIMED_OUT and RETRY_LIMIT results of gate script runner, and I would say that happens several times per day sometimes. however, in my opinion, the necessity for this decision is based not on the fact that the gate script runner job has become more stable, but on the fact that we cannot merge code that has not been properly tested against this job | 00:19 |
|---|---|---|
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #airshipit | 04:33 | |
| *** roman_g has joined #airshipit | 06:38 | |
| *** GoldenBear has quit IRC | 07:43 | |
| *** GoldenBear has joined #airshipit | 07:45 | |
| airship-irc-bot | <se6518> Hello all, please, review proxy settings for airshipctl: https://review.opendev.org/755085 | 15:03 |
| airship-irc-bot | <dwalt> @raliev Thanks for mentioning the timeouts, I hadn't noticed those. The original plan in making the job non-voting was that cores would still only merge code if it has successfully passed the runner job, but we wouldn't have to keep rechecking until the code passed the second run in the gates. Several times for day seems reasonable, but we can always move it back if need be. | 16:39 |
| airship-irc-bot | <raliev> I believe it was a right decision to make it votable because anyways we tried not to merge the code which has -1 mark from gate script runner and I hope failed node results like retry_limit will reduce | 16:44 |
| airship-irc-bot | <se6518> @mattmceuen regarding the question about go<>shell. My point is if we can use the shell script for such cases we should use it. Running container->go->exec->cloudcli looks weird. The only option when we can use this approach is when we don't have enough functionality in SDK for the public cloud providers. | 17:26 |
| *** roman_g has quit IRC | 17:37 | |
| *** born2bake has joined #airshipit | 18:19 | |
| *** roman_g has joined #airshipit | 19:06 | |
| *** roman_g has quit IRC | 19:39 | |
| *** roman_g has joined #airshipit | 19:40 | |
| mattmceuen | @se6518 I think the consensus on the design call was that exec'ing from go was reasonable (in this particular case, not necessarily in general case) -- it certainly gets us to the finish line for bootstrap containers faster, with less effort than integrating with the SDKs | 20:20 |
| mattmceuen | I agree that it is fair to follow that on with the question/discussion of whether it's a better long-term strategy to migrate to the SDKs (where they provide what we need) | 20:21 |
| airship-irc-bot | <mattmceuen> @raliev @kk6740 @ih616h this change: https://review.opendev.org/#/c/752767/22/tools/document/validate_site_docs.sh | 20:22 |
| airship-irc-bot | <mattmceuen> unfortunately broke document validation | 20:22 |
| airship-irc-bot | <mattmceuen> Or rather, defeated it :slightly_smiling_face: | 20:22 |
| airship-irc-bot | <mattmceuen> now that the phase names don't match the directory names, it just skips over validation of most of the phases | 20:23 |
| airship-irc-bot | <mattmceuen> The easy fix would be to make the directory names match the new phase names, but that would get a little redundant too -- e.g. "target/initinfra-target" | 20:24 |
| airship-irc-bot | <mattmceuen> well, now that I'm looking closer, I'm not sure that we even need to know the directory names anymore | 20:25 |
| airship-irc-bot | <mattmceuen> The only gap I see is the if statement (L116) that evaluates whether or not a phase is defined, and only validates it if so -- tracing through the phase plan would be better, but that's not queryable yet, right? | 20:27 |
| airship-irc-bot | <kk6740> well, tbh do we need that validation, it seems very unstable and shaky | 20:27 |
| airship-irc-bot | <mattmceuen> I don't think it has ever been unstable or shaky? | 20:28 |
| airship-irc-bot | <mattmceuen> And yeah, we need it, closest thing we have to linting our manifests | 20:28 |
| airship-irc-bot | <mattmceuen> I found the issue after pushing bad manifests to a patchset twice ;) better to catch locally with the validation script | 20:29 |
| airship-irc-bot | <kk6740> that’s a valid point | 20:29 |
| airship-irc-bot | <kk6740> what i mean, is i am not sure about it’s design. but since we don;t have anything else, i guess we should fix it] | 20:30 |
| airship-irc-bot | <mattmceuen> +1 | 20:31 |
| airship-irc-bot | <mattmceuen> not intended to be the permanent solution, just to get us to `airshipctl document validate` | 20:31 |
| airship-irc-bot | <raliev> it’s fixable either way you proposed, I can do that | 20:32 |
| airship-irc-bot | <kk6740> i think we have somene working on phase plan , that would output phases | 20:32 |
| airship-irc-bot | <mattmceuen> @raliev are we able to query the phase list for a cluster yet? | 20:32 |
| airship-irc-bot | <mattmceuen> phase *plan | 20:32 |
| airship-irc-bot | <kk6740> for manifest u mean? | 20:32 |
| airship-irc-bot | <mattmceuen> yeah | 20:32 |
| airship-irc-bot | <kk6740> we have an issue for that, https://github.com/airshipit/airshipctl/issues/358 | 20:33 |
| airship-irc-bot | <kk6740> Nikarika is working on that i think | 20:33 |
| airship-irc-bot | <mattmceuen> The script needs to loop over some list of phases for each site, for `airshipctl phase apply $phase`, and that'll be different per-site (and per-cluster-in-site) | 20:34 |
| airship-irc-bot | <kk6740> well as of now, i dont think we have command to list phases | 20:34 |
| airship-irc-bot | <kk6740> we have all the libraries to do that | 20:34 |
| airship-irc-bot | <mattmceuen> Yeah, once @niha.twinkle’s change is done, I think that will be the right way to fix the script | 20:34 |
| airship-irc-bot | <mattmceuen> Thanks for volunteering @raliev :slightly_smiling_face: let's keep an eye on the PS and then fix the script after it merges; agree? | 20:35 |
| airship-irc-bot | <raliev> sure :slightly_smiling_face: | 20:35 |
| airship-irc-bot | <mattmceuen> awesome | 20:35 |
| *** roman_g has quit IRC | 21:32 | |
| *** born2bake has quit IRC | 21:45 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!