Earlier  
Posted Nick Remark
#openstack-cyborg - 2021-02-04
02:04:17 xinranwang so let's start
02:04:27 chenke ok
02:04:47 xinranwang #startming
02:05:16 xinranwang #startmeeting openstack_cyborg
02:05:17 openstack Meeting started Thu Feb 4 02:05:16 2021 UTC and is due to finish in 60 minutes. The chair is xinranwang. Information about MeetBot at http://wiki.debian.org/MeetBot.
02:05:18 openstack Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
02:05:20 openstack The meeting name has been set to 'openstack_cyborg'
02:05:36 xinranwang #topic rollcall
02:05:44 xinranwang #info xinranwang
02:05:49 swp20 #info swp20
02:05:49 chenke #info chenke
02:06:12 xinranwang #topic vgpu implementation
02:06:39 xinranwang do you have some thing want bring up about vgpu
02:07:08 chenke yumeng have a new work now.
02:07:19 swp20 nothing from myside
02:07:28 swp20 wait yument response
02:07:35 swp20 yumeng
02:08:46 xinranwang yumeng has left zte now?
02:08:55 chenke yes.
02:09:18 chenke the day before yesterday.
02:09:25 xinranwang oh.
02:09:42 xinranwang so does anyone hand over the vgpu staff
02:10:12 chenke maybe recently, I have no time to update her patch because some office task. sorry.
02:10:44 chenke because of another project task.
02:10:59 xinranwang it's ok
02:12:37 xinranwang is it possible that swp20 can update the patch according to the test report. do you have time to do this?
02:13:42 swp20 ok i will try.
02:14:47 xinranwang swp20: thanks! we will continue review this. hope we can land it in this release. lol
02:15:02 swp20 np
02:15:08 chenke cool.
02:15:17 brinzhang_ hi all
02:15:27 xinranwang hi brinzhang_
02:16:11 chenke hi brin.
02:16:23 xinranwang #topic nic driver
02:16:31 brinzhang_ ^^ pls continue
02:17:11 xinranwang thanks for your connent s on nic driver. I have update it. please check if I answered your question
02:17:19 xinranwang *comment
02:17:20 swp20 i have CR+2 for nic driver
02:17:23 chenke I had check.
02:17:39 chenke about the main code , I din't understand your comment.
02:17:47 brinzhang_ xinranwang: we have concern of the config of enable_drivers in https://review.opendev.org/c/openstack/cyborg/+/758942/7/cyborg.conf.sample#62
02:18:32 brinzhang_ if there is not intel nic that we can write a driver for it
02:18:39 xinranwang I am not able to accese to gerrit now
02:19:01 brinzhang_ but if there is also a new nic for intel and it's not x710, how does it to config?
02:19:29 brinzhang_ xinranwang: nothing, you know what I said
02:20:23 xinranwang it should be in another section. like [x810-static]
02:21:03 xinranwang and nic driver will read this section according to the device id discovered
02:21:10 chenke should we need register it again in cyborg/conf/devices file?
02:24:01 xinranwang chenke: I dont think so
02:24:28 brinzhang_ xinranwang: get your mean, it's ok for me
02:24:33 chenke just config it in cyborg.conf is ok?
02:24:36 brinzhang_ and there is another question physical_device_mappings = physnet1:eth2;eth3
02:24:50 xinranwang we can add x810-static in enabled-nic-types
02:24:52 brinzhang_ does we use ";" instead of ","
02:25:02 brinzhang_ s/does/do
02:25:48 xinranwang yeah. I paste the neutron config format in my reply
02:26:11 xinranwang I prefer to keep the same format as neutron
02:26:23 brinzhang_ IMO, we should use "," to config the options, in nova, cinder they are all use ','
02:26:34 chenke oh. your means we just add x810-static in enabled-nic-types, and then config [x810-static] section is ok, right?
02:26:41 brinzhang_ I think ";" is too old config
02:26:59 xinranwang chenke: yes. and we can add x810-static in enabled-nic-types. it will register implictly
02:27:28 brinzhang_ We should follow the habits of most people
02:27:35 xinranwang brinzhang_: I will check it
02:27:38 chenke good catch brin. ";" and ",",I didn't find it before.
02:28:20 chenke xinranwang, seems “;”  need to ensure about it.
02:28:29 xinranwang what if we have 2 mapping
02:28:32 brinzhang_ xinranwang: thanks
02:29:01 brinzhang_ another thing is the report tests, I think we should maintain it in the docs too. https://wiki.openstack.org/wiki/Cyborg/TestReport/IntelNic
02:29:23 xinranwang eg GTPv1:eth3;eth2
02:29:23 brinzhang_ because we dont have the CI to check the driver
02:29:28 chenke resource_provider_inventory_defaults = allocation_ratio:1.0,min_unit:1,step_size:1,reserved:0
02:29:41 chenke I saw this config in neutron ovs.conf
02:30:07 xinranwang GTPv1:eth3;eth2,GTPV2:eth0;eth1
02:30:43 xinranwang brinzhang_: what docs
02:31:00 brinzhang_ such as the driver matrix
02:31:59 brinzhang_ https://docs.openstack.org/cyborg/latest/reference/support-matrix.html
02:32:15 xinranwang what if we have GTPv1:eth3;eth2,GTPV2:eth0;eth1. you prefer GTPv1:eth3,eth2;GTPV2:eth0,eth1.right?
02:32:21 brinzhang_ what do you think? let maintain this int he driver matrix, is it ok?
02:33:16 xinranwang brinzhang_: driver matrix is ok for me
02:33:27 chenke let's first help xinran ensure the conf.
02:33:48 xinranwang chenke: yeah. thanks haha
02:35:09 xinranwang i see neutron excluded use ; for 1:N mapping
02:35:33 xinranwang *excluded_device
02:35:55 xinranwang that's why I? use ;
02:36:43 xinranwang sorry I use cellphone. so many typos...
02:37:46 brinzhang_ I think that is an exception, I suggest using , to instead of ;
02:38:32 xinranwang if we have 2 mapping, what seperator you prefer
02:38:49 xinranwang GTPv1:eth3,eth2;GTPV2:eth0,eth1.right?
02:38:55 brinzhang_ yes
02:38:58 xinranwang ok
02:39:09 brinzhang_ chenke: is that your concern too?
02:39:34 xinranwang it's ok for me
02:39:52 chenke Yes.You remind me. I didn't see any sample about xinran's situation.
02:40:12 chenke In my opinion. I think GTPv1:eth3,eth2;GTPV2:eth0,eth1 is ok.
02:40:15 xinranwang never mind
02:40:18 xinranwang lol
02:40:46 xinranwang just want got an agreement. the change is small
02:41:05 chenke yes. we are a team. lol.
02:41:21 brinzhang_ yeah, this is the first thing need to be changed
02:41:46 xinranwang so 2 things need improve. the config format and driver matrix doc

Earlier   Later