| kata-irc-bot | <fidencio> @eric.ernst, @mikko.ylinen asked me about the plans, on the Kata Containers side, to start taking advantage of the Sandbox API that was reecently merged. Do you think it's worth bringing this discussion to one of the A/C meetings? | 05:55 |
|---|---|---|
| kata-irc-bot | <liuj97> Last time when reading the containerd core sandbox API, I haven't realized that will affect kata:) | 08:21 |
| kata-irc-bot | <liuj97> Anymore information on this topic? | 08:21 |
| kata-irc-bot | <mikko.ylinen> My understanding is that it's needed to get things like Image Transfer service implemented for CC pods | 08:42 |
| kata-irc-bot | <fidencio> @christophe, @pmores, I've updated https://github.com/kata-containers/kata-containers/pull/4233 addressing your comments. Your thumbs up / approval would be very much welcome :slightly_smiling_face: | 09:46 |
| kata-irc-bot | <liuj97> For that task, we still need enhancements to the CRI part, and may have dependency on containerd 1.7/2.0 arch? | 09:48 |
| kata-irc-bot | <christophe> Done | 09:48 |
| kata-irc-bot | <mikko.ylinen> @liuj97 CRI part as in the spec or containerd's CRI plugin? | 10:15 |
| kata-irc-bot | <fidencio> Thanks! <3 | 10:17 |
| kata-irc-bot | <liuj97> containerd/CRI | 11:18 |
| kata-irc-bot | <fidencio> @christophe, abusing of your presence, if you could take a look at https://github.com/kata-containers/tests/pull/4785 that's the counter-part for installing virtiofsd as part of the CI. After that one gets merged, we can actually start testing on a second commit that I will open against the `kata-containers` repo | 11:26 |
| kata-irc-bot | <fidencio> @gabriela.cervantes.te, it seems the power Ci has been consistently failing due to: ```[cleanup_env.sh:57] INFO: Checks that pods were not left be817d56298fe62f8be2042b3ae46b834ff980524b4c64f367bc276d36d72902 ERROR: 1 pods left and found at /run/vc/sbs``` Shall we make it non-required or wait till Monday so @amulmek1 has a chance to debug it? | 16:30 |
| kata-irc-bot | <gabriela.cervantes.te> @fidencio ok let me do that | 16:38 |
| kata-irc-bot | <fidencio> I will open an issue to have it tracked | 16:39 |
| kata-irc-bot | <eric.ernst> FYI -- I'm going to send a PR, but sharing here since the commit is a bit older, and its something to be cognizant of when doing PR reviews: https://github.com/kata-containers/kata-containers/pull/3923/commits/ff17c756d2cb777ce63af2c9aa5a3dd7dc5e2eba#r872587917 | 16:40 |
| kata-irc-bot | <eric.ernst> I'm looking to add more tests so that way this can be caught in static checks too, but, let's try to protect some of our abstractions. | 16:40 |
| kata-irc-bot | <fidencio> Hmm. Maybe I spoke too soon, as I thought I saw two of the same failures, consecutively. But I cannot find the second. I re-ran the CI and if it passes I will re-mark it as required (sorry for the noise) and move my issue to the tests repo instead. | 16:47 |
| kata-irc-bot | <fidencio> Ack, I will pay more attention to that. | 16:48 |
| kata-irc-bot | <cameron4mcdermott> I have noticed an issue in the stop containers flow where all containers, if the SIGTERM handle was installed or not, would always get sent SIGKILL. It looks to be an issue in `is_signal_handled` when converting from the hex string returned by `SigCgt` to u64. In some cases this hex string is preceded by a separator (tab or space) that causes the conversion to fail with a message like: `failed to parse the str | 16:59 |
| kata-irc-bot | \\t0000000000000000 to hex` . I have opened a PR which should resolve this by checking if the first character is numeric and removing it if not https://github.com/kata-containers/kata-containers/issues/4250 https://github.com/kata-containers/kata-containers/pull/4252 | 16:59 |
| kata-irc-bot | <fidencio> In any way, Eric. It's a super valid head's up on something that directly affects you, but we shouldn't do for the overall quality of the code. | 18:57 |
| kata-irc-bot | <fidencio> And I *really* appreciate the head's up! | 18:57 |
| kata-irc-bot | <fidencio> @eric.ernst, @gabriela.cervantes.te, quick question, do you happen to know if the Ci will explode if I do something like: • PR A on tests (will be merged soon) • PR B, on kata-containers depends on PR A • PR C, on tests, depends on PR B (from kata-containers, that depends on PR A) | 19:41 |
| kata-irc-bot | <eric.ernst> Oh, .... that might rip the fabric of time | 20:48 |
| kata-irc-bot | <fidencio> Cool, thanks @eric.ernst! | 20:55 |
| kata-irc-bot | <fidencio> Folks, https://github.com/kata-containers/kata-containers/pull/4253 and https://github.com/kata-containers/tests/pull/4789 are the last PRs doing the transition from the C version of virtiofsd to the rust version. Right now the transition was only done for x86_64, and on Monday I will open issues with instructions on how to proceed with the transition for the other architectures (and other things that I will notice that I miss, which | 20:59 |
| kata-irc-bot | for sure happened ;-)). | 20:59 |
| kata-irc-bot | <fidencio> CI has been green, including kata-deploy tests with the rust version. | 21:00 |
| kata-irc-bot | <eric.ernst> In case it all explodes, though, it was great working with you Fabiano. | 21:00 |
| kata-irc-bot | <fidencio> I will take that and end my week in a good way :slightly_smiling_face: | 21:01 |
| kata-irc-bot | <fidencio> Likewise, Eric, likewise! | 21:02 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!