| *** egernst has quit IRC | 03:16 | |
| *** egernst_ has joined #kata-dev | 03:16 | |
| *** egernst_ has quit IRC | 03:41 | |
| *** egernst has joined #kata-dev | 03:41 | |
| *** egernst has quit IRC | 04:54 | |
| *** sameo has joined #kata-dev | 05:00 | |
| *** pcaruana has joined #kata-dev | 06:13 | |
| *** fgiudici has joined #kata-dev | 07:11 | |
| *** jodh has joined #kata-dev | 07:16 | |
| *** pcaruana has quit IRC | 07:43 | |
| *** dklyle has quit IRC | 08:01 | |
| *** davidgiluk has joined #kata-dev | 08:05 | |
| *** pcaruana has joined #kata-dev | 08:29 | |
| *** amorenoz has quit IRC | 08:59 | |
| *** khyr0n has quit IRC | 10:16 | |
| *** khyr0n has joined #kata-dev | 10:16 | |
| *** fuentess has joined #kata-dev | 12:58 | |
| *** crobinso has joined #kata-dev | 13:04 | |
| kata-irc-bot | <fidencio> Folks, the AC meeting is at *3PM UTC*, in 1:13 from the time this message was sent! | 13:47 |
|---|---|---|
| kata-irc-bot | <eric.ernst> :slightly_smiling_face: | 14:24 |
| *** dklyle has joined #kata-dev | 14:45 | |
| *** egernst has joined #kata-dev | 15:00 | |
| *** egernst has quit IRC | 15:05 | |
| *** crobinso has quit IRC | 15:06 | |
| kata-irc-bot | <fidencio> This was a short meeting, thanks everyone who joined! | 15:13 |
| kata-irc-bot | <eric.ernst> (sorry i missed it!) | 15:15 |
| kata-irc-bot | <fidencio> Summing up here: • We'll tag a 2.1.0-alpha2 in order to get the tag correctly used; • We agreed on having 2.1.0-rc0 tagged (and branched) on Apr 23rd • We agreed on having 2.1.0 released on May 7th • I'll send an email with a doodle about the community use cases so we can check whether we can have better times (for instance, for folks in AUS & US) for some usecases, if needed. | 15:18 |
| kata-irc-bot | <fidencio> @james.o.hunt, @eric.ernst, I'm slightly lost on how I could trigger the following action on github, for testing purposes: https://github.com/kata-containers/kata-containers/blob/main/.github/workflows/kata-deploy-test.yaml Do you guys know happen to know how to trigger that? Spoiler, `/test-kata-deploy` or `test-kata-deploy` doesn't seem to be he way: https://github.com/kata-containers/kata-containers/pull/1633 | 15:55 |
| kata-irc-bot | :slightly_smiling_face: | 15:55 |
| kata-irc-bot | <eric.ernst> That was the trigger when I added before | 15:56 |
| kata-irc-bot | <eric.ernst> I *think* the action I used for triggers had some side effects of emailing whenver a non-matching trigger was observed, and there was work to adjust. | 15:56 |
| kata-irc-bot | <eric.ernst> TBH, I hadn't looked back at the action workflow since that happened. | 15:56 |
| kata-irc-bot | <james.o.hunt> @fidencio - looks like it's a variant: `/test kata deploy` ? https://github.com/kata-containers/community/wiki/Controlling-the-CI | 15:57 |
| kata-irc-bot | <james.o.hunt> Are those spaces required @eric.ernst? | 15:57 |
| kata-irc-bot | <eric.ernst> maybe /test kata-deploy? | 15:58 |
| kata-irc-bot | <fidencio> I just tried `/test kata-deploy`, seems that no, no change. And that would trigger the whole CI as `/test` itself would trigger the whole CI :slightly_smiling_face: | 15:59 |
| kata-irc-bot | <fidencio> Okay, I guess I found the error! | 16:12 |
| kata-irc-bot | <fidencio> argh, kata-deploy action takes into consideration whatever is in the `main` branch, instead of using the PR as basis :S | 16:46 |
| *** fgiudici has quit IRC | 16:55 | |
| kata-irc-bot | <fidencio> okay, what I'm going to do is removing that check entirely, and re-adding it as part of another PR. With that hopefully we can have it working as expected. | 16:57 |
| *** jodh has quit IRC | 16:59 | |
| *** sameo has quit IRC | 17:00 | |
| kata-irc-bot | <fidencio> I'd appreciate an ack on https://github.com/kata-containers/kata-containers/pull/1635 so I can just move forward with re-adding a functional version of this test | 17:03 |
| *** fuentess has quit IRC | 17:14 | |
| *** sameo has joined #kata-dev | 17:24 | |
| *** egernst has joined #kata-dev | 17:29 | |
| *** egernst has quit IRC | 17:39 | |
| *** egernst has joined #kata-dev | 17:40 | |
| *** crobinso has joined #kata-dev | 17:52 | |
| *** egernst has quit IRC | 17:56 | |
| kata-irc-bot | <eric.ernst> i didn't totally follow why we need to remove, rather than modify the workflow? | 18:59 |
| kata-irc-bot | <eric.ernst> that job isn't designed to test changes to ./src | 19:02 |
| kata-irc-bot | <eric.ernst> it designed to test changes to the actual kata-deploy scripting itself. | 19:02 |
| kata-irc-bot | <eric.ernst> that's why it pulls from main. | 19:02 |
| kata-irc-bot | <eric.ernst> I think we'd be best off _adding_ a test, rather than removing this one. | 19:02 |
| kata-irc-bot | <eric.ernst> The trigger /slash check is likely needing love; other than that I _think_ the test should've been working as planned. | 19:03 |
| kata-irc-bot | <fidencio> @eric.ernst, I'll re-add the test shortly. The problem was that I'd always hit the very same error over and over again, because of a malformed yaml file (mea culpe here!) | 19:20 |
| kata-irc-bot | <fidencio> @eric.ernst, also, I'm super happy to have my mind changed, but I sincerely think we could and should run the full test (build everything + test kata-deploy) on every PR | 19:22 |
| kata-irc-bot | <fidencio> it'd ensure that when time for a release comes, things will be just working | 19:22 |
| kata-irc-bot | <eric.ernst> We could do static builds for each. | 19:28 |
| kata-irc-bot | <eric.ernst> I'm not sure its worth going and starting an AKS cluster for each PR. | 19:28 |
| kata-irc-bot | <eric.ernst> I mean, if our CI runs for over 45 minutes (it does), I guess running yet another 30 minute job doesn't hurt... but... | 19:28 |
| kata-irc-bot | <eric.ernst> Ah, I see -- and yaml changes aren't taken into account here -- we are just testing the container image (dockerfile). | 19:29 |
| *** egernst has joined #kata-dev | 19:29 | |
| kata-irc-bot | <fidencio> our CI already takes over than 45 minutes anyways :S | 19:30 |
| *** davidgiluk has quit IRC | 19:34 | |
| *** egernst has quit IRC | 19:34 | |
| kata-irc-bot | <gmmaharaj> @eric.ernst Just went back and tested and it seems 2.1-alpha0 version of kata works fine running a simple busybox contianer where as running with version 2.0.1, I am hit with this error. ``` Normal Scheduled 13s default-scheduler Successfully assigned default/busybox1 to test-kata Warning FailedCreatePodSandBox 0s (x12 over 12s) kubelet Failed to create pod sandbox: rpc error: code = | 19:50 |
| kata-irc-bot | NotFound desc = failed to create containerd task: not found``` Is there a new release in the 2.0.x series? | 19:50 |
| kata-irc-bot | <gmmaharaj> aah, i see a `2.0.2` tag. Just not a kata-deploy container that goes along with it | 19:53 |
| kata-irc-bot | <gmmaharaj> @fidencio @eric.ernst do you know if kata-deploy action ran for 2.0.2 release? I tried checking if there is a contianer image and i couldn't find it. Could it be a side-effect of the above mentioned issue? | 19:56 |
| kata-irc-bot | <fidencio> I don't think so. I think @bergwolf may not have added the 2.0.2 tag, just that. | 19:57 |
| kata-irc-bot | <fidencio> I'll check with him Tomorrow morning | 19:57 |
| kata-irc-bot | <gmmaharaj> aaah got it. thanks @fidencio. Will keep an eye out for the container tag | 19:58 |
| kata-irc-bot | <fidencio> sorry about that, it's partially caused by my patches /o\ | 19:58 |
| kata-irc-bot | <gmmaharaj> hehe. no worries. ;) as long as we have a way out! | 19:59 |
| kata-irc-bot | <fidencio> @gmmaharaj till last release we're simply tagging and releasing without actually setting a version to pull | 19:59 |
| kata-irc-bot | <fidencio> so users trying to use 2.x, unless they explicitly modified the version, they'd pull 1.x and use it without even noticing /o\ | 19:59 |
| kata-irc-bot | <gmmaharaj> ohh my. | 20:02 |
| *** khyr0n has quit IRC | 20:10 | |
| *** khyr0n has joined #kata-dev | 20:10 | |
| *** sameo has quit IRC | 20:21 | |
| kata-irc-bot | <eric.ernst> I know @fidencio (hope his notifications are snoozed) was looking into issues here. | 20:27 |
| kata-irc-bot | <eric.ernst> 2.0.2 doesn't have artifacts pushed yet -- @bergwolf is working through this. | 20:27 |
| kata-irc-bot | <eric.ernst> I haven't had BW to look at 2.0.1, but surprsied it wouldn't work (it would've had to have passed our sanity tests for kata-deploy on release in order to be published) | 20:27 |
| kata-irc-bot | <gmmaharaj> it could be the version of k8s & containerd maybe? not sure which one. but yeah, 2.1-alpha0 runs good. | 20:38 |
| kata-irc-bot | <gmmaharaj> so something got fixed atleast in my env ;) | 20:38 |
| kata-irc-bot | <fidencio> @eric.ernst, is there some trick for the action to recognise secrets.DOCKER_{USERNAME,PASSWORD}? | 20:51 |
| kata-irc-bot | <fidencio> https://github.com/kata-containers/kata-containers/runs/2281975634?check_suite_focus=true | 20:51 |
| kata-irc-bot | <fidencio> https://github.com/kata-containers/kata-containers/pull/1633/files#diff-fb3fe04e2701915bdfe1f81e9837a0474be5714d63f0cc7a0b1fc9a77020b736R10-R14 | 20:52 |
| kata-irc-bot | <eric.ernst> I'm.... not entirely sure. | 20:53 |
| kata-irc-bot | <eric.ernst> the secrets are tied to the repo: https://github.com/kata-containers/kata-containers/settings/secrets/actions | 20:53 |
| kata-irc-bot | <eric.ernst> are you running from your fork though? | 20:53 |
| kata-irc-bot | <eric.ernst> the secrets have to be from the actual repo. | 20:53 |
| kata-irc-bot | <eric.ernst> It *may* work if you make branch in the repo itself rather than fork. | 20:53 |
| kata-irc-bot | <eric.ernst> but, basically, if you fork the branch, all the secrets don't carry over. | 20:54 |
| kata-irc-bot | <eric.ernst> you'd need equivalents in our repo. | 20:54 |
| kata-irc-bot | <fidencio> that's running on `kata-containers` repo, from my PR, but on `kata-containers` repo | 20:54 |
| kata-irc-bot | <eric.ernst> its running from _your_ fork of kata-containers | 20:56 |
| kata-irc-bot | <eric.ernst> They don't allow that. If i was naughty, I would fork the repo, and write a new workflow in github actions that simply echo's the secrets | 20:56 |
| kata-irc-bot | <eric.ernst> (well, since it will redact the actual secret being printed, I'd shift each character in the string by 1 so i could get the secrets) | 20:57 |
| kata-irc-bot | <fidencio> So, how to test whether actions actually work? :-) | 20:57 |
| kata-irc-bot | <eric.ernst> add your own secrets locally. | 20:57 |
| kata-irc-bot | <eric.ernst> or yolo. i'd need to look at the current 'best practices' | 20:57 |
| kata-irc-bot | <fidencio> Why not YOLO, right? :slightly_smiling_face: | 20:58 |
| kata-irc-bot | <fidencio> the thing is, I have no credentials that will allow me to push to that specific repo | 20:58 |
| kata-irc-bot | <fidencio> s/repo/registry, but you got it | 20:59 |
| kata-irc-bot | <eric.ernst> yeah, you'd have to s/kata-containers/fidencio in the yaml. | 20:59 |
| kata-irc-bot | <eric.ernst> or, again, you could try to make your PR from inside the org's repo instead of your fork. | 20:59 |
| kata-irc-bot | <eric.ernst> I'm not sure what happens there -- only folks who have admin writes can do so, so perhaps its 'less risk' and it'll work. | 21:00 |
| kata-irc-bot | <fidencio> I wonder then, how we'd be testing kata-deploy in general, if PR that would trigger that test would face the same issue that I'm facing | 21:01 |
| kata-irc-bot | <eric.ernst> You mean, how we'd test changes to the workflow yaml? | 21:03 |
| kata-irc-bot | <fidencio> imagine my PR now totally fixes the `test-kata-deploy`, and that we want to have that running on each PR. | 21:03 |
| kata-irc-bot | <fidencio> the next person who opens a PR wouldn't be exactly in the same situation as I am? | 21:04 |
| kata-irc-bot | <fidencio> as their PR wouldn't be coming from inside the org's repo | 21:04 |
| kata-irc-bot | <eric.ernst> if their PR touches the workflow yaml, i think you'd be correct. | 21:07 |
| kata-irc-bot | <eric.ernst> if its a normal PR, all is fine. | 21:08 |
| kata-irc-bot | <eric.ernst> I think. | 21:08 |
| kata-irc-bot | <eric.ernst> can you test your PR by pushing to kata-containers org's repo? | 21:08 |
| kata-irc-bot | <eric.ernst> curious if that works. | 21:08 |
| kata-irc-bot | <fidencio> from what I'm reading here, we can | 21:08 |
| kata-irc-bot | <fidencio> I'm sorry, but I fail to see the point of a test that will only test code that's already merged and / or coming from inside the org. | 21:10 |
| kata-irc-bot | <fidencio> So, as I'd like to see this running one every single PR, I'll try to take the approach of building a local registry as part of the action, and then pushing the image and doing the tests via the local registry | 21:17 |
| kata-irc-bot | <fidencio> at least with this we can ensure the PR is working / will not break what's currently working | 21:18 |
| kata-irc-bot | <fidencio> @eric.ernst, does this sound reasonable? | 21:18 |
| kata-irc-bot | <eric.ernst> You still need secrets to use AKS, etc. | 21:22 |
| kata-irc-bot | <eric.ernst> I think the tests still run for PRs; just you only ever use the workflow that's in the repo. | 21:23 |
| kata-irc-bot | <eric.ernst> aka, i can use that test to verify changes to the kata-deploy dockerfile. | 21:23 |
| kata-irc-bot | <eric.ernst> I just can't expect to modify the test workflow yaml and have _that_ be tested | 21:23 |
| kata-irc-bot | <eric.ernst> Happy to talk in realtime if you want! | 21:24 |
| kata-irc-bot | <fidencio> they'll run, and fail for the PRs | 21:28 |
| *** pcaruana has quit IRC | 21:28 | |
| kata-irc-bot | <fidencio> because we can't access the secrets | 21:29 |
| kata-irc-bot | <eric.ernst> we have an example PR that changes code and not yaml that fails? | 21:29 |
| kata-irc-bot | <fidencio> well, no, because the yaml file was broken before: https://github.com/kata-containers/kata-containers/actions/runs/723175503 | 21:30 |
| kata-irc-bot | <eric.ernst> okay, let's fix the yaml, get it in, _then_ test code changes? | 21:31 |
| kata-irc-bot | <fidencio> okay, let me open a PR specifially for this then | 21:31 |
| kata-irc-bot | <eric.ernst> Last success was 3 mo ago on runtime. | 21:33 |
| kata-irc-bot | <eric.ernst> (which takes ~15 minutes to run) | 21:35 |
| kata-irc-bot | <fidencio> https://github.com/kata-containers/kata-containers/pull/1641 | 21:38 |
| kata-irc-bot | <fidencio> So, let's have this merged, after that https://github.com/kata-containers/kata-containers/pull/1633 can be rebased and the test should pass | 21:40 |
| kata-irc-bot | <fidencio> Okay, calling it a day, we can keep discussing Tomorow. :slightly_smiling_face: | 21:52 |
| *** fuentess has joined #kata-dev | 22:39 | |
| *** egernst has joined #kata-dev | 22:58 | |
| *** egernst has quit IRC | 23:05 | |
| *** fuentess has quit IRC | 23:22 | |
| *** fuentess has joined #kata-dev | 23:32 | |
| *** crobinso has quit IRC | 23:43 | |
| *** fuentess has quit IRC | 23:44 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!