| *** sameo has quit IRC | 00:57 | |
| *** auk_ is now known as auk | 01:46 | |
| *** auk_ has joined #kata-general | 03:45 | |
| *** auk has quit IRC | 03:46 | |
| *** auk_ is now known as auk | 03:55 | |
| *** auk has quit IRC | 07:16 | |
| *** sgarzare has joined #kata-general | 07:51 | |
| *** sameo has joined #kata-general | 08:11 | |
| kata-irc-bot | <durand.didier> Hi there, I just published a repo testing in a fully automated manner replacement of runc by kata-runtime in MicroK8s: https://github.com/didier-durand/microk8s-kata-containers/ The purpose is to make it easier to start for people interested in Kata technology. Its my first post in this group, please let me know if inappropriate. Didier | 08:44 |
|---|---|---|
| *** sgarzare_ has joined #kata-general | 08:47 | |
| *** sgarzare has quit IRC | 08:47 | |
| *** CeeMac has joined #kata-general | 08:55 | |
| *** th0din has joined #kata-general | 10:19 | |
| *** devimc has joined #kata-general | 12:32 | |
| kata-irc-bot | <s257117> Hello @jose.carlos.venegas.m, thank you for your interest. This is the command I used:`fio --filename=test_seq_write --name=seq_write --fallocate=none --ioengine=libaio --iodepth=8 --direct=1 --buffered=0 --invalidate=1 --thread=0 --numjobs=1 --group_reporting=1 --size=4G --filesize=4G --rw=write` . It seems that using kata-emu with 9p as shared fs it is not possible to disable caching at all, so ti performs better than virtiofs, i | 12:49 |
| kata-irc-bot | suppose. | 12:49 |
| *** EricAdamsZNC2 has quit IRC | 13:40 | |
| *** EricAdamsZNC has joined #kata-general | 13:42 | |
| *** sameo has quit IRC | 14:03 | |
| *** devimc has quit IRC | 14:12 | |
| *** devimc has joined #kata-general | 14:15 | |
| *** sameo has joined #kata-general | 15:28 | |
| *** devimc has quit IRC | 15:39 | |
| *** devimc has joined #kata-general | 15:39 | |
| kata-irc-bot | <jose.carlos.venegas.m> @s257117 hey yeah so similar to an fio tests I am tracking , I see direct IO has none effect in host cache for shared fs | 15:53 |
| kata-irc-bot | <jose.carlos.venegas.m> at least if those has support | 15:53 |
| kata-irc-bot | <jose.carlos.venegas.m> as does not have support for cache page | 15:54 |
| kata-irc-bot | <jose.carlos.venegas.m> @durand.didier cool I never tried MicroK8s :P , I wonder if runtimeclass is supported today | 16:02 |
| kata-irc-bot | <jose.carlos.venegas.m> or is it a complete runc -> kata replacement needed | 16:03 |
| *** sgarzare_ has quit IRC | 17:00 | |
| *** pvdp665564 has quit IRC | 17:35 | |
| kata-irc-bot | <s257117> @jose.carlos.venegas.m actually I can't understand what I'm doing wrong. Did some research on the internet and found that most of the benchmarks state that virtiofs should perform better that 9p. Also if I run simple test using dd command i obtain the same bandwidth. | 17:58 |
| kata-irc-bot | <jose.carlos.venegas.m> @s257117 does not use the default 9pfs configuration, it changes a few things like msize | 18:00 |
| kata-irc-bot | <jose.carlos.venegas.m> and also we have one patch that makes 9pfs faster by not using coroutines. | 18:01 |
| kata-irc-bot | <jose.carlos.venegas.m> https://github.com/kata-containers/packaging/blob/stable-1.12/qemu/patches/5.0.x/0001-9p-removing-coroutines-of-9p-to-increase-the-I-O-per.patch | 18:01 |
| kata-irc-bot | <jose.carlos.venegas.m> this is also an improvement that is recently added by | 18:02 |
| kata-irc-bot | <jose.carlos.venegas.m> in this kind of benchmarks depeding the machine I am seeing 9pfs works faster | 18:03 |
| kata-irc-bot | <jose.carlos.venegas.m> https://github.com/jcvenegas/mrunner/tree/master/scripts/bash_workloads | 18:04 |
| kata-irc-bot | <jose.carlos.venegas.m> I am automating that comparison to find more bottlenecks | 18:05 |
| kata-irc-bot | <jose.carlos.venegas.m> let me push my most recent changes | 18:05 |
| kata-irc-bot | <jose.carlos.venegas.m> it uses kata-deploy and builds qemu-virtiofs with the most recent changes from Vivek, I think soon will add those patches to Kata | 18:06 |
| kata-irc-bot | <jose.carlos.venegas.m> this is the equivalent patch | 18:07 |
| kata-irc-bot | <jose.carlos.venegas.m> https://github.com/rhvgoyal/qemu/commit/823dd12f7a2f393632c86b98e8ee380044282d12 | 18:07 |
| kata-irc-bot | <jose.carlos.venegas.m> about runc I see dropping not use direct IO helps but still see qemu being faster | 18:08 |
| kata-irc-bot | <s257117> thank you for the hints. I will try your scripts and check the result | 18:29 |
| *** auk has joined #kata-general | 21:19 | |
| *** devimc has quit IRC | 22:03 | |
| *** pvdp665564 has joined #kata-general | 22:07 | |
| *** th0din has quit IRC | 22:21 | |
| *** th0din has joined #kata-general | 22:24 | |
| *** sameo has quit IRC | 23:52 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!