| *** zerocoolback has joined #kata-dev | 00:25 | |
| *** bkslsh has joined #kata-dev | 00:34 | |
| *** zerocoolback has quit IRC | 00:46 | |
| *** zerocoolback has joined #kata-dev | 00:46 | |
| *** zerocoolback has quit IRC | 00:50 | |
| *** marst has quit IRC | 00:51 | |
| *** lamego has joined #kata-dev | 00:56 | |
| *** marst has joined #kata-dev | 01:05 | |
| *** lamego has left #kata-dev | 01:05 | |
| *** bkslsh has quit IRC | 01:14 | |
| *** zerocoolback has joined #kata-dev | 01:59 | |
| *** dlw has joined #kata-dev | 02:30 | |
| *** zerocoolback has quit IRC | 03:06 | |
| *** zerocoolback has joined #kata-dev | 03:40 | |
| *** lpetrut has joined #kata-dev | 03:50 | |
| *** sjas has joined #kata-dev | 04:09 | |
| *** sjas_ has quit IRC | 04:12 | |
| *** zerocoolback has quit IRC | 04:14 | |
| *** zerocoolback has joined #kata-dev | 04:39 | |
| *** lpetrut has quit IRC | 04:39 | |
| *** lpetrut has joined #kata-dev | 04:42 | |
| *** zerocoolback has quit IRC | 04:45 | |
| *** zerocoolback has joined #kata-dev | 04:47 | |
| *** lpetrut has quit IRC | 04:54 | |
| *** fdegir has quit IRC | 05:16 | |
| *** din has quit IRC | 05:16 | |
| *** fdegir has joined #kata-dev | 05:35 | |
| *** din has joined #kata-dev | 05:35 | |
| *** core has quit IRC | 05:37 | |
| *** core has joined #kata-dev | 05:37 | |
| *** zerocoolback has quit IRC | 06:18 | |
| *** zerocoolback has joined #kata-dev | 06:18 | |
| *** zerocoolback has quit IRC | 06:18 | |
| *** zerocoolback has joined #kata-dev | 06:19 | |
| *** zerocoolback has joined #kata-dev | 06:20 | |
| *** zerocoolback has quit IRC | 06:20 | |
| *** zerocoolback has joined #kata-dev | 06:20 | |
| *** zerocoolback has quit IRC | 06:21 | |
| *** dlw has quit IRC | 06:21 | |
| *** zerocoolback has joined #kata-dev | 06:21 | |
| *** dlw has joined #kata-dev | 06:21 | |
| *** zerocoolback has quit IRC | 06:21 | |
| *** lpetrut has joined #kata-dev | 06:22 | |
| *** jugs has quit IRC | 06:41 | |
| *** jugs has joined #kata-dev | 06:42 | |
| *** dklyle has quit IRC | 07:35 | |
| *** sameo has quit IRC | 07:49 | |
| *** davidgiluk has joined #kata-dev | 08:03 | |
| *** gwhaley has joined #kata-dev | 08:06 | |
| *** sameo has joined #kata-dev | 08:22 | |
| *** dlw has quit IRC | 09:10 | |
| *** dlw has joined #kata-dev | 09:10 | |
| *** dlw1 has joined #kata-dev | 10:13 | |
| *** dlw has quit IRC | 10:13 | |
| *** dlw1 is now known as dlw | 10:13 | |
| *** gwhaley has quit IRC | 10:58 | |
| *** dlw has quit IRC | 11:08 | |
| *** changcheng has joined #kata-dev | 11:21 | |
| *** JerryLiu has quit IRC | 11:21 | |
| *** sameo has quit IRC | 11:24 | |
| *** changcheng has quit IRC | 11:43 | |
| *** changcheng has joined #kata-dev | 11:46 | |
| *** zerocoolback has joined #kata-dev | 11:50 | |
| *** fuentess has joined #kata-dev | 11:53 | |
| *** zerocoolback has quit IRC | 11:56 | |
| *** zerocool_ has joined #kata-dev | 11:56 | |
| *** devimc has joined #kata-dev | 11:57 | |
| *** fuentess1 has joined #kata-dev | 12:07 | |
| *** fuentess1 has quit IRC | 12:08 | |
| *** gwhaley has joined #kata-dev | 12:14 | |
| * davidgiluk is curious where the connection between DisableModern on virtio devices and nested came from | 12:15 | |
| gwhaley | hi @davidgiluk: iirc, there was a bug/feature that caused 'fast' to not work in the nested situation. iirc, it was reported and fixed upstream (kernel) - but I presume it may have not trickled through all the distros yet. | 12:21 |
|---|---|---|
| gwhaley | ref: https://github.com/intel/govmm/pull/11 but sadly it does not have much detail :-( | 12:21 |
| gwhaley | https://github.com/containers/virtcontainers/pull/664 | 12:22 |
| davidgiluk | thanks! | 12:28 |
| davidgiluk | hmm that references 663 that mentions azure VMs | 12:30 |
| gwhaley | I don't remember where we (first?) saw it - mcastelino might, but he is not here right now .... devimc may remember some details? | 12:35 |
| devimc | @davidgiluk, @gwhaley sorry, what was the question? | 12:36 |
| devimc | why DisableModern ? | 12:36 |
| gwhaley | hi devimc - why did we have to ... ^^ yeah, for nested - was an upstream kernel limitation iirc - do you remember any details, and on what nested service etc.? | 12:37 |
| devimc | technical details? no, but DisableModern=true is to make happy azure VMs | 12:38 |
| davidgiluk | ok, so it sounds just like an azure special - useful one to know | 12:41 |
| gwhaley | give me a moment - I might have a link... | 12:41 |
| gwhaley | possibly fixed with https://patchwork.kernel.org/patch/10330065/ | 12:42 |
| gwhaley | grabbed from https://bugs.launchpad.net/qemu/+bug/1636217 | 12:42 |
| openstack | Launchpad bug 1636217 in QEMU "qemu-kvm 2.7 does not boot kvm VMs with virtio on top of VMware ESX" [Undecided,Fix released] | 12:42 |
| gwhaley | that EPT reference sounds familiar | 12:43 |
| davidgiluk | oh ok; that's fun | 12:44 |
| kata-irc-bot1 | <bergwolf> Do you have refs for the upstream bug/feature? I can't find it in github log | 12:55 |
| kata-irc-bot1 | <bergwolf> oops, just saw the whole conversation. sorry for the noise... | 12:57 |
| *** mestery has quit IRC | 13:18 | |
| *** mestery has joined #kata-dev | 13:18 | |
| *** dklyle has joined #kata-dev | 13:50 | |
| *** sameo has joined #kata-dev | 13:51 | |
| *** lamego has joined #kata-dev | 13:57 | |
| *** dklyle has quit IRC | 14:00 | |
| *** eernst has joined #kata-dev | 14:14 | |
| *** annabelleB has joined #kata-dev | 14:35 | |
| *** lpetrut has quit IRC | 14:47 | |
| *** sgordon has quit IRC | 14:51 | |
| clarkb | salvador.fuentes: it looked like unittest addition was working but then crio failed. Are you happy enough with that result or should I avoid merging the job update? | 15:03 |
| kata-irc-bot1 | <salvador.fuentes> clarkb: I think we can merge, I am seeing issues with the checkcommits tool (it is reporting that there are zero commits to verify). I think it is mainly because of the way zuul is managing the git branches. | 15:06 |
| clarkb | is it trying to look at diffs if on unmerged code? | 15:07 |
| *** lpetrut has joined #kata-dev | 15:16 | |
| gwhaley | clarkb: checkcommits tries to locate the patches that are in the PR it is checking so it can do sanity checks on them (formatting, format etc.) - so it tries to identify what is different between HEAD / master and the branch to be checked. | 15:22 |
| gwhaley | we had to do a little smarts as different CIs set things up differently in the build branches. Maybe we will have to educate it about ZUUL as well | 15:23 |
| clarkb | couldn't you just do sanity checking on the whole since it includes the PR patches? | 15:23 |
| gwhaley | Travis was... the 'worst' so far... | 15:23 |
| gwhaley | clarkb - yes, and no. I don't think we can guarantee that all historical code or commits fits with our current requirements for instance | 15:23 |
| clarkb | well you wouldn't step through the tree, just that the current commit meets the standards | 15:24 |
| clarkb | so if you got everything up to the standard and enforced thsi check you just check the current checkout each time | 15:24 |
| gwhaley | right - but it is not just one commit - it is the series of commits that make up the PR - they can be multi-commit | 15:24 |
| clarkb | I understand but it is a DAG and the tip of the PR will include the other commits as well | 15:25 |
| clarkb | and if you care about intermediate states the PR model isn't very good :/ | 15:25 |
| gwhaley | Ah, right - that is how Zuul works isn't it iirc - it is a pile of all current pending commits? (sorry if pile is not quite the right phrase...) | 15:26 |
| gwhaley | yeah, we don't enforce that bisects work in our codebase btw - as long as the final commit in a PR series works then we are happy. We do try to have some standards though ;-) | 15:26 |
| gwhaley | hmm, so, maybe right now we might have to disable checkcommits in Zuul whilst we stare at it and the Zuul workflow to figure out if anything 'clever' can be done | 15:27 |
| clarkb | gwhaley: there are sort of two related things here. THe first is PR workflows (on their own) are really bad for enforcing the type of thing you describe because intermediate states aren't really enforceable. Gerrit's discrete change model is better (imo) as it forces you to consider each commit in series | 15:27 |
| clarkb | the other is that yes Zuul tries to give you speculative future states to test against because code bases (at least with git) move forward only | 15:28 |
| clarkb | What you are really interested in testing is the commit that will be created when you "merge" to master | 15:28 |
| clarkb | not necesarily the state a commit was pushed in | 15:28 |
| gwhaley | clarkb - yes, but, checkcommits is really more of a 'style checker' - so it is only really checking your commits are 'well formed' https://github.com/kata-containers/tests/tree/master/cmd/checkcommits#overview | 15:29 |
| *** annabelleB has quit IRC | 15:29 | |
| gwhaley | we use it in conjunction with (to fill some gaps of) 'pullapprove'. | 15:29 |
| gwhaley | it checks the commit messages, not the code :-) | 15:30 |
| * gwhaley spies stefanha... | 15:30 | |
| clarkb | In this case a PR to kata-containers/proxy proposed to the master branch would look like Master_0 <- PR_A <- PR_B. If while reviewing that PR some other PR merges to master what Zuul will test is Master_1 <- PR_A' <- PR_B' and in the test instance master will be checked out to PR_B' | 15:30 |
| *** annabelleB has joined #kata-dev | 15:30 | |
| clarkb | gwhaley: aha! that is what I was missing you are looking at commits themselves not their contents | 15:31 |
| gwhaley | Hi stefanha - have you seen https://github.com/kata-containers/agent/pull/263 - I wonder if you have any thoughts/insight - as you were <cough> implicated on the original qemu/glib hang up diagnosis, and this feels very similar | 15:31 |
| gwhaley | clarkb: :-) | 15:31 |
| clarkb | PR_B' on master is not what upstream code repo looks like, but it is what it would look like if that PR merged in the future | 15:31 |
| clarkb | and we test that rather than PR_B which is now a state that can never materialize | 15:31 |
| clarkb | (I actually think github does some of this auto rebasing for you too now, but forget the details) | 15:32 |
| *** lpetrut has quit IRC | 15:37 | |
| clarkb | gwhaley: for this commit check to function properly it needs to know what the merge target is and what sha1 is already merged to that target? | 15:38 |
| gwhaley | we really need @jodh on this, but he is out until next week. What it needs is to be able to figure out which commits come from the PR under test. I think it currently does that by asking git to tell it the difference | 15:40 |
| gwhaley | between the current branch and the master - but, we can adapt it if we need for Zuul if we have to I think | 15:41 |
| clarkb | I'm trying to confirm this, but my memory is that while 'master' is the speculative state there is a refs/origin/master (or similar) that has the target branches existing state. I think it may be as simple as git log $BRANCH..refs/origin/$BRANCH (but will figure out those details | 15:42 |
| *** devimc has quit IRC | 15:43 | |
| *** lpetrut has joined #kata-dev | 15:55 | |
| clarkb | https://zuul-ci.org/docs/zuul/user/jobs.html#git-repositories seems to confirm it | 15:59 |
| *** devimc has joined #kata-dev | 16:01 | |
| gwhaley | cool - we probably need to open an Issue on the tests repo for https://github.com/kata-containers/tests/tree/master/cmd/checkcommits to gather details - are you up for opening that clarkb with the Zuul details? | 16:02 |
| clarkb | gwhaley: yup I can put that together in a bit. | 16:02 |
| gwhaley | cc @fuentess :-) | 16:03 |
| *** zerocool_ has quit IRC | 16:13 | |
| *** zerocoolback has joined #kata-dev | 16:14 | |
| *** zerocoolback has quit IRC | 16:14 | |
| *** lpetrut has quit IRC | 16:17 | |
| *** sameo has quit IRC | 16:35 | |
| *** sameo has joined #kata-dev | 17:12 | |
| *** gwhaley has quit IRC | 17:31 | |
| *** annabelleB has quit IRC | 17:39 | |
| clarkb | https://github.com/kata-containers/tests/issues/511 filed for the above discussion | 17:46 |
| *** annabelleB has joined #kata-dev | 17:46 | |
| *** dklyle has joined #kata-dev | 17:55 | |
| *** dklyle has quit IRC | 18:01 | |
| *** dklyle has joined #kata-dev | 18:27 | |
| *** lpetrut has joined #kata-dev | 18:28 | |
| *** eernst has quit IRC | 18:38 | |
| *** libregeekingkid has joined #kata-dev | 18:39 | |
| *** annabelleB has quit IRC | 19:19 | |
| *** dklyle has quit IRC | 19:23 | |
| *** annabelleB has joined #kata-dev | 19:24 | |
| *** annabelleB has quit IRC | 19:34 | |
| *** davidgiluk has quit IRC | 19:37 | |
| *** annabelleB has joined #kata-dev | 19:41 | |
| *** annabelleB has quit IRC | 19:51 | |
| *** devimc has quit IRC | 20:03 | |
| *** eernst has joined #kata-dev | 20:17 | |
| *** dklyle has joined #kata-dev | 20:39 | |
| *** lpetrut has quit IRC | 20:48 | |
| *** dklyle has quit IRC | 21:42 | |
| *** dklyle has joined #kata-dev | 21:45 | |
| *** david-lyle has joined #kata-dev | 21:53 | |
| *** dklyle has quit IRC | 21:54 | |
| *** david-lyle has quit IRC | 22:31 | |
| *** libregeekingkid has quit IRC | 22:32 | |
| *** eernst has quit IRC | 22:49 | |
| *** annabelleB has joined #kata-dev | 22:54 | |
| *** annabelleB has quit IRC | 22:58 | |
| *** annabelleB has joined #kata-dev | 23:05 | |
| *** dklyle has joined #kata-dev | 23:09 | |
| *** eernst has joined #kata-dev | 23:30 | |
| *** annabelleB has quit IRC | 23:33 | |
| *** annabelleB has joined #kata-dev | 23:44 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!