| *** swp20 has joined #openstack-cyborg | 02:32 | |
| *** Yumeng has joined #openstack-cyborg | 03:02 | |
| swp20 | hi Yumeng | 03:02 |
|---|---|---|
| Yumeng | hi wenping | 03:03 |
| *** xinranwang has joined #openstack-cyborg | 03:06 | |
| xinranwang | Hi all | 03:07 |
| Yumeng | hi xinranwang | 03:07 |
| swp20 | the question about un-hotplug GPU is because the image is not available for cirros, when i use centos, it's ok for re-hotplug. | 03:07 |
| swp20 | hi xinranwang | 03:07 |
| Yumeng | swp20: what's the question? | 03:08 |
| swp20 | i asked you on wechat | 03:09 |
| swp20 | un-hotplug not success with qemu driver in use. | 03:09 |
| Yumeng | let's start the meeting first :) | 03:09 |
| Yumeng | #startmeeting openstack-cyborg | 03:09 |
| openstack | Meeting started Thu Nov 26 03:09:37 2020 UTC and is due to finish in 60 minutes. The chair is Yumeng. Information about MeetBot at http://wiki.debian.org/MeetBot. | 03:09 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 03:09 |
| *** openstack changes topic to " (Meeting topic: openstack-cyborg)" | 03:09 | |
| openstack | The meeting name has been set to 'openstack_cyborg' | 03:09 |
| Yumeng | #topic Roll call | 03:09 |
| *** openstack changes topic to "Roll call (Meeting topic: openstack-cyborg)" | 03:09 | |
| Yumeng | #info Yumeng | 03:09 |
| xinranwang | #info xinranwang | 03:10 |
| swp20 | #info swp20 | 03:10 |
| Yumeng | #topic Agenda | 03:10 |
| *** openstack changes topic to "Agenda (Meeting topic: openstack-cyborg)" | 03:10 | |
| Yumeng | # topic vgpu | 03:11 |
| Yumeng | #topic vgpu | 03:11 |
| *** openstack changes topic to "vgpu (Meeting topic: openstack-cyborg)" | 03:11 | |
| Yumeng | swp20: pls continue | 03:13 |
| Yumeng | are you saying the detach failure in hotplug is because of cirros mssing? | 03:14 |
| swp20 | yeah, the cirros vm process crash when detach the gpu device | 03:14 |
| swp20 | the un-hotplug is not real success in fact. | 03:15 |
| swp20 | cirros image is not support and centos is well. | 03:15 |
| Yumeng | so vgpu hotplug is not supported in cirros but supported in centos, right? | 03:18 |
| swp20 | i am not sure. | 03:18 |
| swp20 | i means re-hotplug | 03:18 |
| Yumeng | ok | 03:19 |
| Yumeng | did you find out why un-hotplug is not successful? | 03:19 |
| swp20 | attach, detach and reattach | 03:19 |
| swp20 | i search the vm log | 03:20 |
| swp20 | there is process crash problem. | 03:21 |
| Yumeng | is it a occasional case or it crash every time? | 03:23 |
| swp20 | it's high probability | 03:23 |
| Yumeng | ok | 03:23 |
| Yumeng | Has this crash ever happend in Centos? | 03:24 |
| swp20 | hasn't met yet. | 03:25 |
| Yumeng | ok. got taht. | 03:25 |
| swp20 | cool | 03:25 |
| Yumeng | looks like hotplug is image sensitive. | 03:26 |
| Yumeng | Thanks wenping for the sharing | 03:26 |
| swp20 | maybe the driver is important. | 03:26 |
| Yumeng | do you mean nvidia virtualization driver? | 03:27 |
| swp20 | no, i means the driver in image | 03:27 |
| swp20 | gpu is not support well for cirros | 03:28 |
| swp20 | include vgpu | 03:28 |
| swp20 | you can test for vgpu about detach by 'virsh detach-device' | 03:28 |
| Yumeng | yes, the VFIO mdev driver is very important. nvidia virtualization driver version must be well match the image version | 03:29 |
| Yumeng | ok. will try when I got time | 03:29 |
| swp20 | cool | 03:30 |
| Yumeng | I also have a vGPU issue to discuss with you | 03:30 |
| Yumeng | about the vGPU support | 03:30 |
| swp20 | yep | 03:30 |
| swp20 | i think in the time of bind arq is better | 03:30 |
| swp20 | to create mdev | 03:31 |
| swp20 | attach_handle is too early | 03:31 |
| swp20 | and maintain task is heavy | 03:32 |
| Yumeng | yes, I also think so. | 03:32 |
| swp20 | so let's confirm this. | 03:33 |
| Yumeng | xinranwang what do you think? | 03:34 |
| Yumeng | Sylvain prefer create mdev in generate attach_handle. sean and gibi is fine with either | 03:35 |
| Yumeng | from my perspective, I also prefer creating medv when arq bind | 03:36 |
| xinranwang | if gpu's type is determined, the max number of vfs is also determined, right? | 03:37 |
| Yumeng | yes | 03:37 |
| Yumeng | but if it is changed, we need to delete all the created ones and create new ones | 03:38 |
| xinranwang | if we do not create mdev at attach_handle generation step, how many vfs should we report? | 03:38 |
| Yumeng | even if they were never used. | 03:38 |
| Yumeng | xinranwang: the maximum number | 03:40 |
| Yumeng | in the inventory, we always report the maximum number | 03:40 |
| xinranwang | ok, got it. | 03:40 |
| xinranwang | it seems create mdev during binding is more efficient. we just create it when we use it. | 03:41 |
| Yumeng | yes, that's also how I mentioned in nova spec. | 03:42 |
| xinranwang | does mdev creation spend much time? | 03:42 |
| Yumeng | not much. | 03:42 |
| xinranwang | will it fail in some cases? | 03:43 |
| Yumeng | I tested in my env, but it was not a big number of VMs. create mdev is very fast | 03:44 |
| Yumeng | but not sure what's the results when VM is a large number | 03:44 |
| xinranwang | mdev creation is a serial task, i think. | 03:45 |
| xinranwang | anyway, i think at binding step is more efficient, if there is no obvious gap. | 03:46 |
| Yumeng | creation failure is at very Low frequency. | 03:46 |
| Yumeng | hasn't met yet | 03:46 |
| Yumeng | xinranwang: ok. cool | 03:47 |
| Yumeng | So we agreed on at binding step. | 03:47 |
| Yumeng | I will go back to sync with nova guys | 03:47 |
| Yumeng | ok. nothing else from side. | 03:47 |
| Yumeng | Is ther anything else you guys what to mention? | 03:48 |
| xinranwang | nothing from my side | 03:49 |
| Yumeng | ok~ | 03:51 |
| Yumeng | lunch time~~ | 03:51 |
| xinranwang | lol | 03:51 |
| xinranwang | bon appetit | 03:51 |
| Yumeng | haha | 03:52 |
| Yumeng | so let's wrap up today's meeting | 03:52 |
| Yumeng | #endmeeting | 03:52 |
| *** openstack changes topic to "Pending patches (Meeting topic: openstack-cyborg)" | 03:52 | |
| openstack | Meeting ended Thu Nov 26 03:52:41 2020 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 03:52 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/openstack_cyborg/2020/openstack_cyborg.2020-11-26-03.09.html | 03:52 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/openstack_cyborg/2020/openstack_cyborg.2020-11-26-03.09.txt | 03:52 |
| Yumeng | bye | 03:52 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/openstack_cyborg/2020/openstack_cyborg.2020-11-26-03.09.log.html | 03:52 |
| xinranwang | bye | 03:52 |
| *** swp20 has quit IRC | 04:15 | |
| *** xinranwang has quit IRC | 06:06 | |
| openstackgerrit | Xinran WANG proposed openstack/cyborg master: refresh device profile link in API doc https://review.opendev.org/c/openstack/cyborg/+/764281 | 07:27 |
| *** swp20 has joined #openstack-cyborg | 07:46 | |
| openstackgerrit | Xinran WANG proposed openstack/cyborg master: refresh device profile link in API doc https://review.opendev.org/c/openstack/cyborg/+/764281 | 08:20 |
| openstackgerrit | YumengBao proposed openstack/cyborg-specs master: Cyborg NVIDIA GPU Driver support vGPU management https://review.opendev.org/c/openstack/cyborg-specs/+/758925 | 09:35 |
| *** swp20 has quit IRC | 13:35 | |
| *** swp20 has joined #openstack-cyborg | 13:36 | |
| *** songwenping_ has joined #openstack-cyborg | 15:34 | |
| *** swp20 has quit IRC | 15:37 | |
| *** Yumeng has quit IRC | 17:12 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!