| *** tmhoang has quit IRC | 00:06 | |
| *** changcheng has quit IRC | 03:35 | |
| *** eernst has joined #kata-dev | 04:42 | |
| kata-irc-bot | <hejianet> hi all, does docker build use kata-runtime? | 06:56 |
|---|---|---|
| kata-irc-bot | <hejianet> when I run "docker build ", there is no qemu process launched | 06:56 |
| kata-irc-bot | <eric.ernst> depends on how you have docker configured. | 06:59 |
| kata-irc-bot | <eric.ernst> if it is using kata-runtime by default AFAIU it will ty to. | 06:59 |
| kata-irc-bot | <eric.ernst> I *think* there are some limitations on how docker build works where you could see issues when using Kata. | 07:00 |
| kata-irc-bot | <eric.ernst> Its been a while since I've tried personally, though, so YMMV. | 07:00 |
| *** eernst has quit IRC | 07:00 | |
| kata-irc-bot | <hejianet> ok, thanks ~~ | 07:10 |
| *** tmhoang has joined #kata-dev | 07:50 | |
| *** igordc has quit IRC | 08:00 | |
| *** sgarzare has joined #kata-dev | 08:01 | |
| *** jodh has joined #kata-dev | 08:36 | |
| *** gwhaley has joined #kata-dev | 09:03 | |
| *** fiddletwix has quit IRC | 09:07 | |
| *** davidgiluk has joined #kata-dev | 09:10 | |
| *** sameo has joined #kata-dev | 09:23 | |
| kata-irc-bot | <graham.whaley> Hi @salvador.fuentes @gabriela.cervantes.te - what's the current status with the CI migrations from ubuntu 16.04 to 18.04? Are we moving all the x86_64 QA CI jobs over? In which case, I think we may have some cleaning up to do still? For instance, the packaging job is still called 16.04, but is doing 18.04 builds, which is a little confusing ;) http://jenkins.katacontainers.io/job/kata-containers-packaging-ubuntu-16-04-PR/ | 10:05 |
| *** gwhaley has quit IRC | 12:05 | |
| *** gwhaley has joined #kata-dev | 13:13 | |
| *** changcheng has joined #kata-dev | 13:28 | |
| *** devimc has joined #kata-dev | 13:36 | |
| *** sgarzare_ has joined #kata-dev | 13:54 | |
| *** changcheng has quit IRC | 13:56 | |
| *** sgarzare has quit IRC | 13:56 | |
| *** devimc has quit IRC | 15:44 | |
| *** devimc has joined #kata-dev | 15:45 | |
| kata-irc-bot | <gabriela.cervantes.te> @graham.whaley fixed | 15:49 |
| *** tmhoang has quit IRC | 16:15 | |
| *** sameo_ has joined #kata-dev | 16:16 | |
| *** sameo has quit IRC | 16:18 | |
| *** devimc_ has joined #kata-dev | 16:23 | |
| *** devimc_ has quit IRC | 16:23 | |
| *** sgarzare_ has quit IRC | 16:52 | |
| *** devimc has quit IRC | 17:04 | |
| kata-irc-bot | <manohar.r.castelino> @davidgiluk ping | 17:10 |
| davidgiluk | manohar: Hi | 17:12 |
| kata-irc-bot | <manohar.r.castelino> @davidgiluk sebastian and I were testing virtio-fs with DAX enabled | 17:13 |
| kata-irc-bot | <manohar.r.castelino> we saw some strange issues when mmaping a file larger than the cache size | 17:14 |
| davidgiluk | ooh fun; how did it break? | 17:14 |
| kata-irc-bot | <manohar.r.castelino> or if the total amount of memory mmaped excedded the cache size | 17:14 |
| kata-irc-bot | <manohar.r.castelino> we say the VM side call just hang | 17:14 |
| davidgiluk | hmm | 17:14 |
| kata-irc-bot | <manohar.r.castelino> we did not debug it further.. but will do so | 17:14 |
| kata-irc-bot | <manohar.r.castelino> this happens for an explict mmap in the VM | 17:15 |
| davidgiluk | manohar: So the theory is that it should do unmap's and then reallocate - we have seen that work OK but do occasionally trip up nasties | 17:15 |
| davidgiluk | manohar: the kernel works in I think 2MB chunks, sending removemappings to free up space before it does the next mappings | 17:16 |
| kata-irc-bot | <manohar.r.castelino> yes we saw the 2MB boundaries | 17:17 |
| kata-irc-bot | <manohar.r.castelino> we created a VM with cache size 8 MB | 17:17 |
| kata-irc-bot | <manohar.r.castelino> then had three files on the host side 2MB, 4MB and 6MB | 17:17 |
| kata-irc-bot | <manohar.r.castelino> when we mmaped 4MB and 6MB we saw the hang | 17:17 |
| davidgiluk | hehe 8MB - we've never gone anywhere near that low; we tend to use a few GB :-) | 17:17 |
| kata-irc-bot | <manohar.r.castelino> let us reproduct this with a simple test case and send out a note | 17:18 |
| davidgiluk | yes please, please include vivek on that, he knows the kernel map/unmap code best | 17:18 |
| davidgiluk | also, please try a more generous size, it might be something that only happens on tiny mppings | 17:18 |
| *** tmhoang has joined #kata-dev | 17:28 | |
| *** devimc has joined #kata-dev | 18:02 | |
| *** jodh has quit IRC | 18:04 | |
| *** gwhaley has quit IRC | 18:05 | |
| *** igordc has joined #kata-dev | 18:09 | |
| *** tmhoang has quit IRC | 18:37 | |
| *** tmhoang has joined #kata-dev | 19:22 | |
| *** tmhoang has quit IRC | 19:22 | |
| *** sameo_ has quit IRC | 19:23 | |
| *** tmhoang has joined #kata-dev | 19:24 | |
| kata-irc-bot | <manohar.r.castelino> We would like to add a new repo to kata to support a LVM based snapshotter for containerd https://github.com/kata-containers/runtime/pull/1294 | 19:30 |
| kata-irc-bot | <manohar.r.castelino> Sorry wrong link | 19:31 |
| kata-irc-bot | <manohar.r.castelino> https://github.com/kata-containers/runtime/issues/1303 | 19:31 |
| kata-irc-bot | <manohar.r.castelino> @gmmaharaj has an initial implementation that works with both kata and runc. We would like to move the development of this snapshotter to the kata organization so that we get feeback from the broader kata community. | 19:32 |
| *** davidgiluk has quit IRC | 20:22 | |
| *** igordc has quit IRC | 21:09 | |
| *** igordc has joined #kata-dev | 21:09 | |
| *** eernst has joined #kata-dev | 21:48 | |
| *** eernst has quit IRC | 21:52 | |
| *** fiddletwix has joined #kata-dev | 21:59 | |
| *** devimc has quit IRC | 22:05 | |
| *** sameo_ has joined #kata-dev | 23:19 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!