Friday, 2020-05-15

*** crobinso has quit IRC00:07
*** vgoyal has quit IRC00:39
*** sameo has joined #kata-dev05:18
*** dklyle has quit IRC06:27
*** jugs has quit IRC06:39
*** jugs has joined #kata-dev06:40
*** jodh has joined #kata-dev06:42
*** sgarzare has joined #kata-dev07:11
*** davidgiluk has joined #kata-dev08:06
*** gwhaley has joined #kata-dev08:06
*** sameo has quit IRC08:19
*** sameo has joined #kata-dev08:31
*** crobinso has joined #kata-dev11:38
*** devimc has joined #kata-dev11:53
*** vgoyal has joined #kata-dev12:42
*** pcaruana has quit IRC13:58
*** devimc has quit IRC14:08
*** devimc has joined #kata-dev14:09
*** pcaruana has joined #kata-dev14:11
kata-irc-bot<fidencio> @afrosi hey/ping. out of curiosity, the recommended sharedfs to be used with kata on s390x is 9p, is it?14:36
kata-irc-bot<afrosi> yes, we don't have virtiofs support on Z14:37
kata-irc-bot<fidencio> cool, thanks!14:37
kata-irc-bot<afrosi> yw14:37
kata-irc-bot<afrosi> I think we will meet on Monday :slightly_smiling_face:14:37
*** dklyle has joined #kata-dev14:41
kata-irc-bot<eric.ernst> @afrosi - is there a technical blocker for this, or just enabling for this? In Kata, or the VMM, or?14:54
kata-irc-bot<fidencio> I'd guess in the hypervisor code, but either Alice or David could ellaborate more on that15:12
kata-irc-bot<fidencio> @eric.ernst takung advantage you're around, how do you estimate the right values to be used with pod overhead?15:13
kata-irc-bot<eric.ernst> “it depends”15:13
kata-irc-bot<eric.ernst> I’ve been looking at it as the overhead of just running the pod, not necessarily doing a particular workload.15:14
kata-irc-bot<eric.ernst> If you are running a large amount of file I/O, for example, your virtio back ends will be doing a decent amount of work, CPU wise.  You can give a ‘swag’ for this.15:15
kata-irc-bot<eric.ernst> I’ve been mostly concerned around stability, which I see as memory.15:15
kata-irc-bot<eric.ernst> So, just run the VM.  Check the usage at pod cgroup versus sum of containers; that’s the base value I’ve been using.15:16
kata-irc-bot<eric.ernst> As usage deviates, if your system is using horizontal or vertical autoscaling, it’ll take care of the cpu portion.15:16
kata-irc-bot<eric.ernst> beyond the initial swag15:17
kata-irc-bot<fidencio> I see. Yeah, that's a really good start!15:17
kata-irc-bot<fidencio> Thanks for the pointers!15:18
kata-irc-bot<eric.ernst> There are some details here you can checkout: https://kubernetes.io/docs/concepts/configuration/pod-overhead/15:55
*** sgarzare has quit IRC16:44
*** jodh has quit IRC17:03
*** gwhaley has quit IRC17:03
*** DeniseD has joined #kata-dev18:51
*** DeniseD has left #kata-dev19:00
*** DeniseD has joined #kata-dev19:02
*** DeniseD has quit IRC19:22
*** DeniseD has joined #kata-dev19:22
*** davidgiluk has quit IRC20:00
*** devimc has quit IRC20:44
*** crobinso has quit IRC20:53
*** DeniseD has quit IRC21:25
*** vgoyal has quit IRC21:26

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!