| opendevreview | Merged opendev/system-config master: Setup ML for FLOSS Mentorship Group https://review.opendev.org/c/opendev/system-config/+/901610 | 01:08 |
|---|---|---|
| frickler | hmm, flaky connectivity to review.o.o once again it seems. both v4+v6 affected | 07:26 |
| tonyb | Hmmm No Route to host | 07:28 |
| tonyb | The Instance is up and hasn't rebooted so this time it might actually be networking this time | 07:33 |
| tonyb | ``` 17. AS174 38.122.103.106 0.0% 1 254.0 254.0 254.0 254.0 0.0 | 07:34 |
| tonyb | 18. AS33028 review02.opendev.org 0.0% 1 264.0 264.0 264.0 264.0 0.0 | 07:34 |
| tonyb | ``` | 07:34 |
| tonyb | Looks okay | 07:34 |
| frickler | yes, the issue was come and go, not 100% down, can confirm it is looking more stable again now | 07:55 |
| tonyb | okay. I'll be around for a while if you loose access again. | 08:16 |
| jcapitao[m] | hello folks, can we go ahead with https://review.opendev.org/c/openstack/project-config/+/912904 ? | 08:24 |
| opendevreview | Joel Capitao proposed openstack/project-config master: Implement periodic job to update Packstack constraints https://review.opendev.org/c/openstack/project-config/+/912904 | 11:29 |
| Clark[m] | re 912904 my first impression is that shouldn't be in openstack/project-config at all which would necessitate using another Gerrit account. In general though the accounts have no special privileges and only push code that must be reviewed anyway so really the question is about job organization not Gerrit accounts I think. | 13:34 |
| fungi | i think the reason it's in project-config is because that's where the secret is defined for the proposal-bot account? | 13:38 |
| fungi | otherwise i would agree | 13:38 |
| fungi | my main question for now is whether the proposal-bot account is meant to be shared by different projects (it's currently used by both openstack and opendev jobs), or whether every project is expected to have their own account for pushing patches to gerrit | 13:40 |
| Clark[m] | In my head it's more that we shouldn't maintain x/packstack synchronization code in openstack project config. And ultimately that requires a different Gerrit account | 13:42 |
| Clark[m] | In the case of opendev this is part of the legacy of the pre zuulv3 days. It's not a indication of correctness | 13:43 |
| frickler | well the current bot is called "OpenStack Proposal Bot" and I would like that name to stay https://review.opendev.org/q/owner:proposal-bot. moving more stuff from the openstack tenant to opendev seems also like a good idea if someone is looking for a task ;) | 13:43 |
| fungi | yeah, i mean, a simple adjustment there would be to just drop the "openstack" from the display name, the username is already just proposal-bot | 13:45 |
| fungi | but that's only if we intend for it to be used by other projects | 13:46 |
| Clark[m] | I still think that is the wrong question though. If you intend to use the account for any project that implies sharing all the code in one spot like project config. And I don't think we should do that. Therefore different accounts should be used | 13:47 |
| Clark[m] | Because we shouldn't be asked to review packstack synchronization code | 13:47 |
| fungi | yeah, that makes sense. so for openstack translation jobs which propose patches for many different repos, would each one have their own copy of the job and their own account to push with, or would it be put somewhere central (and if so, does that mean a separate trusted config repo) | 13:54 |
| Clark[m] | I think openstack can manage it's translation jobs centrally in openstack project config | 13:55 |
| Clark[m] | Basically openstack project config reviewers shouldn't need to understand packstack puppet needs. But openstack translation jobs are in scope. And eventually if we finish up the opendev tenant migration we'll correct the flaw with those jobs | 13:58 |
| opendevreview | Clark Boylan proposed opendev/system-config master: Rotate clarkbs ssh key https://review.opendev.org/c/opendev/system-config/+/914895 | 14:45 |
| clarkb | fungi: I noticed that git config's gitreview.username config option doesn't seem to be documented in git-review docs. Am I missing it or should I try to add something? | 14:46 |
| frickler | clarkb: it is the first item in the CONFIGURATION section of the manpage? | 14:49 |
| clarkb | oh its not in the web docs then? | 14:53 |
| clarkb | fwiw installing from pypi means you don't get manpages installed... | 14:53 |
| fungi | you do, you just need to add it to your manpath or have a symlink to it in your manpath | 14:54 |
| fungi | or you can directly view it with an invocation like `man ~/lib/git-review/share/man/man1/git-review.1` | 14:55 |
| fungi | but yes, trying to keep things in sync between the manpage and rst (and remembering to ask change authors to update those when they propose a new feature or whatever) is quite a pain | 14:56 |
| clarkb | right there are ways to view it but git review --help and `man git-review` don't work without intervention so I fallback to the regular docs | 14:56 |
| fungi | well, my point was installing from pypi does install the manpage, just not somewhere than man will find them | 14:57 |
| fungi | it sticks it in your venv at share/man/man1/git-review.1 | 14:57 |
| clarkb | got it | 14:58 |
| fungi | just like installing from pypi doesn't install the executable in your default path, so you have to do extra things to make it so | 14:58 |
| fungi | in my case i symlink from ~/bin/ into the bindir of the venv, but there are quite a few possible approaches | 14:59 |
| frickler | shall we just include the man page in the web docs, then? | 15:10 |
| clarkb | does sphinx have a man page render option? I like that idea | 15:11 |
| fungi | i've usually done it the other way around: write rst to embed in the web docs and generate the manpage from the same document | 15:12 |
| fungi | it's also far more convenient to edit rst that *roff | 15:13 |
| clarkb | true I always have to lookup those directives when I look at the raw manpage file in that repo | 15:13 |
| frickler | yes, seems that's what sphinx is made to do https://www.sphinx-doc.org/en/master/usage/configuration.html#options-for-manual-page-output | 15:14 |
| fungi | another option is to make the cli help more thorough, and then feed --help output into both | 15:14 |
| fungi | which could be done in a multi-stage build by first generating rst from --help (with a template wrapper to add the rest of the content that isn't suitable for help output), and generate the manpage content from there | 15:16 |
| fungi | that shrinks the duplication down from 3 to 1 copy of all that information | 15:16 |
| fungi | and in the case of a git subcommand, it also means that the output people get from `git review --help` (which prints the manpage) and `git-review --help` (which shows the argparse utilization/help fields) will have more consistent content even if they're still not identical | 15:18 |
| frickler | oh, I never noticed that difference, nice | 15:20 |
| clarkb | git review -h prints the built in help too fwiw | 15:21 |
| clarkb | its only the --help flag that git overrides | 15:21 |
| fungi | oh wow, i hadn't noticed that nuance | 15:22 |
| frickler | ah, /me is lazy and only uses -h, that may explain not noticing ;) | 15:24 |
| frickler | fyi I'll skip the meeting today, need some more time to digest the dst change | 15:54 |
| fungi | clarkb: do you still want me to chair the meeting, or do you think you'll be set up in time? | 15:57 |
| clarkb | fungi: I think it would be best if you run it | 15:57 |
| fungi | happy to, just making sure! | 15:57 |
| clarkb | I'm still working through my backlog of things to update and with other meetings today I want ot make sure I give myself enough time for that as tomorrow is the release and I have other paperwork stuff to get done | 15:58 |
| clarkb | and for my own sanity I think I need to go outside and ride my bike in a big circle after meetings today | 15:58 |
| SvenKieske | meh, when I want to check some visuals on docs previews I get "Web Listing Disabled" e.g. https://410c0463c0087166dadd-c3ec680ab787b0ae7fbdeb20a072dd12.ssl.cf2.rackcdn.com/750996/3/check/openstack-tox-docs/ff2c6b9/docs/ | 17:43 |
| SvenKieske | is that a known issue? | 17:44 |
| SvenKieske | this is the job in question: https://zuul.opendev.org/t/openstack/build/ff2c6b9780ed4a19895905d3e710230e | 17:44 |
| clarkb | how did you map between the two? | 17:46 |
| SvenKieske | the error seems to be "The owner of this web site has disabled web listing. " and "The owner of this web site has set X-Container-Meta-Web-Index: index.html. However, this file is not found." maybe it's only the index.html missing? | 17:47 |
| clarkb | because none of the logs are currently available so I don't think it is just that the index.html is missing | 17:47 |
| SvenKieske | in the Build, I clicked "Artifacts" tab and then "Docs preview site" | 17:47 |
| SvenKieske | mhm, true, no logs | 17:48 |
| clarkb | ah ok | 17:48 |
| SvenKieske | I'll look at the build history | 17:48 |
| fungi | is it more than a month old? | 17:48 |
| clarkb | oh yup thats exactly it | 17:48 |
| SvenKieske | ah yes, it's actually from january | 17:48 |
| clarkb | I was just looking at other logs from rackcdn from today and it all looks happy | 17:48 |
| clarkb | the problem is those files have been deleted after the 30 day timeout | 17:48 |
| SvenKieske | sure, sorry! I didn't check that first.. | 17:49 |
| fungi | everything that's uploaded from builds gets a 30-day expiration set in swift | 17:49 |
| SvenKieske | I'll quickly do the visual doc test locally, no need to bother the CI with that | 17:49 |
| fungi | `tox -re docs` in a checkout of the change (cherry-picked or merged to the branch if you want to be as close as possible to what zuul will build) and then point a browser at the build/html dir that results | 17:50 |
| fungi | that's what i usually do anyway | 17:51 |
| SvenKieske | yeah; it's not my own change, just was doing a code review and wanted to check something visually and didn't notice the last CI run was quite some time ago :D | 17:52 |
| fungi | git remote update && git checkout origin/master && git review -x 750996 && tox ... | 17:53 |
| *** persia is now known as Guest4708 | 19:30 | |
| corvus1 | okay, i'm looking at nb01 now to see what's going on with the up-and-to-the-right disk graph | 21:29 |
| fungi | i prefer to look at it as down-and-to-the-left | 22:15 |
| corvus1 | fungi: clarkb it is an issue related to the new feature and it has broken builds; remote: https://review.opendev.org/c/zuul/nodepool/+/914928 Fix deletion of intermediate image builds [NEW] should fix it | 22:30 |
| clarkb | looking now | 22:36 |
| opendevreview | Merged openstack/project-config master: Temporarily remove release docs semaphores https://review.opendev.org/c/openstack/project-config/+/914688 | 22:39 |
| tonyb | Random question. If I created a flavor that was tied to always be baremetal machines deployed via nova+ironic. What do you think the issues would be using those in conjunction with nodepool for zuul ci resources? | 22:40 |
| clarkb | tonyb: we might need special nodepool images to boot on them (I'm not sure if our VM images would just work(tm)). Otherwise I think the main concerns woudl be on the cloud side ensuring that the hardware isn't owned (could be mitigated via rules around what jobs can run there and ironic host cleanup steps between boots) | 22:49 |
| corvus1 | also cloud diversification is a consideration (similar to arm or nested virt) | 22:51 |
| tonyb | clarkb: Thanks. Those are good points. I was more worried about the turn around time with baremetal being "slow" | 22:51 |
| tonyb | corvus1: Ahhhh yes I can see that being an issue too. | 22:52 |
| clarkb | I'm less concerned about turn around time because jobs selecting those labels should expect that | 22:53 |
| clarkb | it may still be an issue but one that comes with the tradeoff of baremetal access | 22:53 |
| corvus1 | i've deleted a whole bunch of stuff from dib_tmp on nb01 and nb02, including ancient build tmp files, just to avoid potential issues from running out of space while we await the fix | 23:12 |
| clarkb | thanks! | 23:21 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!