| *** pcaruana has joined #kata-dev | 05:07 | |
| *** pcaruana has quit IRC | 05:35 | |
| *** kata-irc-bot has quit IRC | 06:49 | |
| *** pcaruana has joined #kata-dev | 06:51 | |
| *** kata-irc-bot has joined #kata-dev | 06:51 | |
| *** sgarzare has joined #kata-dev | 07:14 | |
| *** sameo has joined #kata-dev | 07:21 | |
| *** eernst has joined #kata-dev | 07:28 | |
| *** eernst has quit IRC | 07:56 | |
| *** davidgiluk has joined #kata-dev | 07:59 | |
| *** jodh has joined #kata-dev | 08:00 | |
| *** jugs1 has quit IRC | 10:27 | |
| *** jugs1 has joined #kata-dev | 10:28 | |
| *** jodh has quit IRC | 10:36 | |
| *** jugs1 has quit IRC | 12:30 | |
| *** jugs1 has joined #kata-dev | 12:31 | |
| *** devimc has joined #kata-dev | 13:09 | |
| *** david-lyle is now known as dklyle | 13:33 | |
| *** fuentess has joined #kata-dev | 13:47 | |
| *** dklyle has quit IRC | 14:02 | |
| *** dklyle has joined #kata-dev | 14:04 | |
| *** fuentess has quit IRC | 14:30 | |
| *** fuentess has joined #kata-dev | 14:34 | |
| *** pcaruana has quit IRC | 14:39 | |
| *** pcaruana has joined #kata-dev | 15:01 | |
| *** pcaruana has quit IRC | 15:26 | |
| *** eernst has joined #kata-dev | 15:29 | |
| *** devimc has quit IRC | 15:53 | |
| *** devimc has joined #kata-dev | 15:53 | |
| brtknr | stefanha: looks like the issue persists but I believe its more to do with interaction with beegfs | 15:55 |
|---|---|---|
| brtknr | https://seashells.io/p/UvpvPj3B | 16:01 |
| brtknr | the same config works find when its using 9pfs | 16:01 |
| *** jodh has joined #kata-dev | 16:01 | |
| brtknr | the same config works find when its using qemu+9p | 16:02 |
| brtknr | these failures are when I use nemu+virtiofs | 16:02 |
| brtknr | these failures are when I use nemu+virtiofs+beegfs | 16:04 |
| stefanha | brtknr: "fio: io_u error on file data/fio_write/beegfs-randwrite-1-vtwtr/3.dat: Invalid argument: write offset=8754351104, buflen=256" looks like spurious error from virtio_fs | 16:04 |
| brtknr | it works when its nemu+virtiofs+ceph | 16:04 |
| stefanha | brtknr: Have you tweaked the mount command-line in any way? | 16:05 |
| *** eernst has quit IRC | 16:05 | |
| brtknr | no its line by line deployment from packaging/kata-deploy | 16:05 |
| brtknr | no its verbatim deployment from packaging/kata-deploy | 16:05 |
| brtknr | stefanha: from here: https://github.com/kata-containers/packaging/tree/master/kata-deploy | 16:06 |
| stefanha | brtknr: The debug output from virtiofsd would be very useful. There is a chance that this error is coming from virtio_fs.ko inside the guest, but it's more likely that virtiofsd is hitting the EINVAL error on the host side. | 16:08 |
| stefanha | brtknr: If you enable kata-runtime debug in /etc/kata-containers/configuration.toml, then virtiofsd will write debug output to stderr. stderr goes to nowhere but you can still capture it with strace. | 16:09 |
| stefanha | brtknr: It's possible to tell strace to capture the full string for write() syscalls, and also to filter only on write() syscalls. That way the log should contain little noise and mostly just the virtiofsd stderr log write() syscalls. | 16:10 |
| stefanha | brtknr: Does what I'm saying make sense? I haven't tried it out and don't know the exact strace command-line without looking through the man page, but I think this would be a good way to collect debug output. | 16:10 |
| brtknr | Ok I’ll try that tomorrow, afk now | 16:11 |
| stefanha | brtknr: Thanks! | 16:13 |
| *** canyounot has joined #kata-dev | 16:19 | |
| *** igordc has joined #kata-dev | 16:20 | |
| *** dklyle has quit IRC | 16:30 | |
| *** david-lyle has joined #kata-dev | 16:30 | |
| *** david-lyle is now known as dklyle | 16:31 | |
| *** pcaruana has joined #kata-dev | 16:40 | |
| *** sgarzare has quit IRC | 16:49 | |
| *** jodh has quit IRC | 17:05 | |
| *** noahm has joined #kata-dev | 17:19 | |
| *** sameo has quit IRC | 18:00 | |
| *** igordc has quit IRC | 18:19 | |
| *** davidgiluk has quit IRC | 19:18 | |
| *** igordc has joined #kata-dev | 20:03 | |
| *** pcaruana has quit IRC | 20:15 | |
| *** canyounot is now known as canyounot|gone | 20:56 | |
| *** canyounot|gone has quit IRC | 20:56 | |
| *** devimc has quit IRC | 21:25 | |
| *** sameo has joined #kata-dev | 22:16 | |
| *** igordc has quit IRC | 22:36 | |
| *** fuentess has quit IRC | 22:46 | |
| brtknr | stefanha: has 1.9.3 kata nemu runtime reverted to using virtio-9p by default | 22:52 |
| kata-irc-bot | <eric.ernst> 1.8.3 * | 22:54 |
| kata-irc-bot | <eric.ernst> If you want latest virito-fs, there's a version of QEMU that includes it. | 22:55 |
| kata-irc-bot | <eric.ernst> look at 1.9.0-rc0 | 22:55 |
| kata-irc-bot | <eric.ernst> brtknr ^^ | 22:55 |
| brtknr | @eric.ernst why was this decision taken | 22:56 |
| kata-irc-bot | <eric.ernst> we are deprecating NEMU, ultiamtely. | 22:57 |
| kata-irc-bot | <eric.ernst> there were a bunch of patch changes required to support virtio-fs in QEMU which would need to be ported to NEMU. | 22:57 |
| kata-irc-bot | <eric.ernst> Latest QEMU's kconfig changes are great, and reduce the need for something like NEMU. | 22:57 |
| kata-irc-bot | <eric.ernst> team that was doing NEMU is now focusing on a rust-vmm based VMM. | 22:57 |
| kata-irc-bot | <eric.ernst> I can pull up message on ML re this. Are you using NEMU now? Or what vmm? | 22:58 |
| brtknr | @eric-ernst ah okay thanks, im attempting to run qemu with virtiofs but hititng this: Failed create pod sandbox: rpc error: code = Unknown desc = failed to get sandbox runtime: no runtime for "kata-qemu-virtiofs" is configured | 23:01 |
| kata-irc-bot | <eric.ernst> docker? | 23:01 |
| kata-irc-bot | <eric.ernst> and, kata-runtime --version? | 23:01 |
| brtknr | kata-runtime : 1.9.0-rc0 | 23:03 |
| brtknr | commit : a4b5a565ff589e7203a10d2c3ae1a37cf91c0d9d | 23:03 |
| brtknr | OCI specs: 1.0.1-dev | 23:03 |
| brtknr | with kubernetes | 23:03 |
| kata-irc-bot | <eric.ernst> cool. | 23:03 |
| brtknr | ive applied the runtimeclass | 23:03 |
| kata-irc-bot | <eric.ernst> packages or tarball? | 23:03 |
| brtknr | ummm | 23:03 |
| brtknr | I built the Dockerfile on the master branch | 23:04 |
| kata-irc-bot | <eric.ernst> experimental is available via kata-deploy only (though i'm not sure if the container image for rc0 has been published yet -- thehy are working on it this moment) | 23:04 |
| brtknr | hmm but im able to run nemu and qemu with master branch | 23:05 |
| brtknr | just not qemu-virtiofs | 23:05 |
| brtknr | @eric.ernst^ | 23:06 |
| kata-irc-bot | <eric.ernst> there's a specific kernel that qemu-virtiofs needs | 23:07 |
| kata-irc-bot | <eric.ernst> and QEMU, for that matter. | 23:07 |
| kata-irc-bot | <eric.ernst> both of those shuld be in release tarball. | 23:07 |
| kata-irc-bot | <eric.ernst> sorry for the thrash - the release is stll being finzlied, and we should spell this out in release notes, etc. | 23:09 |
| brtknr | @eric.ernst okay will i be able to use nemu with 1.8.3 in the mean time? | 23:10 |
| brtknr | with virtiofsd support? | 23:10 |
| brtknr | stefanha: this is my journalctl with debug enabled https://seashells.io/v/gAwVRyjX | 23:25 |
| brtknr | this may be easier to search https://seashells.io/p/gAwVRyjX | 23:25 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!