kata-irc-bot | <xwlpt> I try to run the perf at the vm. One curious issue I found that | 01:02 |
---|---|---|
kata-irc-bot | <xwlpt> | 01:02 |
kata-irc-bot | <xwlpt> update_cfs_group takes many cpu usage. | 01:03 |
kata-irc-bot | <xwlpt> | 01:03 |
kata-irc-bot | <xwlpt> The cpu usage is free. | 01:03 |
kata-irc-bot | <xwlpt> ```bash-4.2# uname -r 4.19.86``` | 01:04 |
kata-irc-bot | <xwlpt> The is my kernel version | 01:04 |
kata-irc-bot | <eric.ernst> Hey @xwlpt — can you clarify what exaclty you’re profiling (from when to when) ? | 03:13 |
kata-irc-bot | <xwlpt> Just running perf top | 03:31 |
kata-irc-bot | <xwlpt> @eric.ernst Do you tried this before. Maybe you have more context for this. | 03:33 |
*** sameo has joined #kata-general | 04:13 | |
*** bpradipt has joined #kata-general | 04:42 | |
*** sameo has quit IRC | 06:15 | |
*** sameo has joined #kata-general | 07:38 | |
*** sgarzare has joined #kata-general | 07:46 | |
*** sameo has quit IRC | 07:48 | |
*** sameo has joined #kata-general | 07:57 | |
*** sameo_ has joined #kata-general | 08:01 | |
*** sameo has quit IRC | 08:04 | |
*** sgarzare_ has joined #kata-general | 09:55 | |
*** sgarzare has quit IRC | 09:57 | |
*** sgarzare_ has quit IRC | 10:07 | |
*** sgarzare_ has joined #kata-general | 10:08 | |
*** sgarzare_ has quit IRC | 10:15 | |
*** sgarzare has joined #kata-general | 10:30 | |
*** sgarzare has quit IRC | 11:57 | |
*** devimc has joined #kata-general | 12:07 | |
*** hashar has joined #kata-general | 12:18 | |
*** sgarzare has joined #kata-general | 12:24 | |
*** sgarzare_ has joined #kata-general | 13:19 | |
*** sgarzare has quit IRC | 13:19 | |
*** devimc has quit IRC | 13:28 | |
*** devimc has joined #kata-general | 13:28 | |
*** CeeMac has joined #kata-general | 13:44 | |
kata-irc-bot | <xwlpt> Hi, can someone did a perf for the qemu process? I did it, and there something wired. I’d wonder to know this is only happens at my test env, or it is common. | 13:57 |
kata-irc-bot | <xwlpt> | 13:57 |
kata-irc-bot | <xwlpt> | 13:58 |
*** hashar has quit IRC | 15:38 | |
*** sgarzare_ has quit IRC | 16:25 | |
*** sameo_ has quit IRC | 17:26 | |
*** bpradipt has quit IRC | 18:27 | |
*** hashar has joined #kata-general | 18:29 | |
*** devimc has quit IRC | 19:15 | |
*** devimc has joined #kata-general | 19:25 | |
*** CeeMac has quit IRC | 20:04 | |
*** tobberydberg_ has joined #kata-general | 20:12 | |
*** th0din_ has joined #kata-general | 20:13 | |
*** irclogbot_3 has quit IRC | 20:14 | |
*** CeeMac has joined #kata-general | 20:15 | |
*** tobberydberg has quit IRC | 20:18 | |
*** th0din has quit IRC | 20:18 | |
*** sjas has quit IRC | 20:18 | |
*** irclogbot_2 has joined #kata-general | 20:20 | |
*** sjas has joined #kata-general | 20:23 | |
*** devimc has quit IRC | 21:10 | |
*** hashar has quit IRC | 21:52 | |
kata-irc-bot | <eric.ernst> @xwlpt I don’t have any feedback just yet, but am interested in digging into this, and in particular I would be curious to see the same done for cloud hypervisor (@chen.bo heads up). I wonder if LiuBin’s observability PR will be helpful here, as well, to gain insights (heads up @bergwolf). | 23:03 |
*** CeeMac has quit IRC | 23:34 | |
kata-irc-bot | <chen.bo> Hi @eric.ernst @xwlpt, thanks for prompting the issues. Right, profiling and performance measurement are very interesting topics to cover for both clh and its use in kata. We definitely want to dig into more, and hopefully in the near future. We will keep you posted. | 23:37 |
kata-irc-bot | <chen.bo> @xwlpt If you can share the steps and commands you used on profiling qemu and kata-runtime. I can do a similar experiment and provide with you a reference if that's useful to you. :slightly_smiling_face: | 23:38 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!