| kata-irc-bot | <fidencio> <here>, welcome to the release week! I'd like to ask each and every one of you to let me know, as a reply to this thread, what you want to have merged before 2.5.0-alpha1 is cut, and also to work on the things you think that should be part of 2.4.0 release (but were not yet backported). Please, either raise a PR and send the PR link to this thread, or just let me know in this thread and I can take care of raising the PR (and I will | 07:00 |
|---|---|---|
| kata-irc-bot | bother people for a review on those). | 07:00 |
| kata-irc-bot | <gkurz> @fidencio Hi ! The clh-devmapper check seems to be consistently failing for a bunch of PRs in the test repo... ```00:07:42 not ok 2 Empty dir volume when FSGroup is specified with non-root container 00:07:50 # (from function `assert_equal' in file k8s-empty-dirs.bats, line 16, 00:07:50 # in test file k8s-empty-dirs.bats, line 61) 00:07:50 # `assert_equal "1001" "$uid"' failed 00:07:50 # INFO: k8s configured to use runtimeclass 00:07:50 | 09:21 |
| kata-irc-bot | # Image is up to date for sha256:a6487be0b50adf014d15508876d7be8185398e140055cc8f87c341e92b2ea98e 00:07:50 # pod/sharevol-kata created 00:07:50 # sharevol-kata 0/2 Completed 0 3s 00:07:50 # expected: 1001, got: 00:07:50 # Name: sharevol-kata 00:07:50 # Namespace: default 00:07:50 # Priority: 0 00:07:50 # Node: ubuntu1804-azuree97540/10.0.0.14 00:07:50 # Start Time: Fri, 22 Apr 2022 22:07:47 +0000 00:07:50 | 09:21 |
| kata-irc-bot | # Labels: <none> 00:07:50 # Annotations: <none> 00:07:50 # Status: Failed``` Any idea ? | 09:21 |
| kata-irc-bot | <fidencio> Can you point me to a PR where it's failing? | 09:45 |
| kata-irc-bot | <gkurz> @fidencio https://github.com/kata-containers/tests/pull/4655 | 09:46 |
| kata-irc-bot | <fidencio> Can you open an issue for that? | 09:48 |
| kata-irc-bot | <fidencio> Fow now we re-run the CI but I, unfortunately, have zero time this week to debug this issue. | 09:49 |
| kata-irc-bot | <gkurz> @fidencio https://github.com/kata-containers/tests/issues/4728 | 10:03 |
| kata-irc-bot | <fidencio> @bergwolf, @samuel.ortiz, I need some help with devmapper + Cloud Hypervisor behaviour with related to disks. :thread: | 10:50 |
| kata-irc-bot | <fidencio> So, my understanding is that with devmapper we would: 1. Boot the rootfs using `virtio-pmem` 2. Hotplug the DM device to the VM, as virtio-blk 3. Use the DM / virtio-blk for every volume added / used with the container, but not the rootfs (as that's already up and running, using `virtio-pmem`) | 10:53 |
| kata-irc-bot | <fidencio> Is this correct or I am totally off? | 10:54 |
| kata-irc-bot | <fidencio> In case I'm correct, is there any reason to *not* expose virtio-blk as an option for the rootfs? | 11:22 |
| kata-irc-bot | <bergwolf> I'm not familiar with CLH. Waiting for @samuel.ortiz to address it. | 11:27 |
| kata-irc-bot | <fidencio> Thanks, @bergwolf! | 11:28 |
| kata-irc-bot | <fidencio> Hmm. It seems the rootfs is then remounted inside the to using the virtio-blk, as dmesg shows: ```[ 0.556053] virtio-pci 0000:00:05.0: enabling device (0000 -> 0002) [ 0.563979] virtio_blk virtio4: [vda] 8388608 512-byte logical blocks (4.29 GB/4.00 GiB) [ 0.577891] EXT4-fs (vda): mounted filesystem with ordered data mode. Opts: . Quota mode: disabled.``` And then mount shows me: ```/dev/vda on | 12:36 |
| kata-irc-bot | /run/kata-containers/shared/containers/test type ext4 (rw,relatime,stripe=16) /dev/vda on /run/kata-containers/test/rootfs type ext4 (rw,relatime,stripe=16)``` | 12:36 |
| kata-irc-bot | <samuel.ortiz> @fidencio virtio-blk could be used instead of virtio-pmem. I think the latter is a better option. | 13:01 |
| kata-irc-bot | <samuel.ortiz> And yes, all container images and volumes are then mounted as virtio-blk disks, afair. | 13:01 |
| kata-irc-bot | <samuel.ortiz> It’s a looong time ago… | 13:02 |
| kata-irc-bot | <fidencio> pmem is indeed a better option, but I am trying to understand better how the whole dance of having this setup with devmapper works. | 13:02 |
| kata-irc-bot | <fidencio> And how to, then, i/o limit those. | 13:02 |
| kata-irc-bot | <fidencio> We have no rate limiter option for pmem, due to the pmem nature | 13:03 |
| kata-irc-bot | <fidencio> <here>: We don't have any topics for the A/C meeting tomorrow, and this is a release week ... so, it's been a busy one. I'd propose we cancel this week's meeting, and meet again on April 3rd! | 16:10 |
| kata-irc-bot | <gkurz> s/April/May but makes sense :slightly_smiling_face: | 16:17 |
| kata-irc-bot | <fidencio> Fixed, wow, this year is passing way too fast. :slightly_smiling_face: | 16:25 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!