13:00:32 #startmeeting sriov 13:00:35 Meeting started Tue Aug 23 13:00:32 2016 UTC and is due to finish in 60 minutes. The chair is moshele. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:38 The meeting name has been set to 'sriov' 13:00:51 hi 13:00:52 hi 13:01:04 hi 13:01:41 #topic Improving SR-IOV/PCI Passthrough CI 13:02:02 so lenny is on PTO so I don't have any update here 13:02:29 we merged Fixed manager.py to support multinode test on vnic_port - https://review.openstack.org/#/c/335447/ 13:02:34 so this is good 13:02:47 I didn't see any reviews on the migration tests 13:03:01 lbeliveau: are you still working on them? 13:03:12 yeah, I still need to address lenny's comment, will work on them end of this week 13:03:32 cool 13:04:19 also we will try to do PF passthrough CI 13:04:33 we may get the red hat help on this :) 13:04:58 cool, done with mellanox ci ? 13:05:04 yes 13:05:35 we tested a bit the PF pass through and the basic functionality is working 13:06:08 #topic Patches for Subteam Review 13:06:43 so we have 2 critical patches for migration-revert that need core reviews 13:07:03 Update binding:profile for SR-IOV ports on resize-revert - https://review.openstack.org/#/c/347444/ and Fix drop_move_claim() on revert resize - https://review.openstack.org/#/c/349060 13:07:36 I told mriedem that these reviews are stuck in last nova team meeting 13:07:46 lbeliveau: maybe jaypipes can review again Fix drop_move_claim() on revert resize - https://review.openstack.org/#/c/349060 13:07:54 yes 13:08:09 late hi 13:08:13 lbeliveau: so I know his on PTO let bug dansmith :) 13:08:18 hi 13:08:34 wznoinsk: hi 13:08:46 Sorry 13:09:12 I will try to look today. Remind me if I don't :) 13:09:32 dansmith: sure, thanks :) 13:10:16 lbeliveau: regarding this pci: in free_device method compare devices according to pci_device id - https://review.openstack.org/#/c/347558/ 13:10:40 lbeliveau: did you see my respond to jaypipes ? 13:11:10 moshele: reviewing 349060 now.. 13:11:20 moshele: I haven't, didn't see that in my emails, will look at it today 13:12:17 perhaps we need to make sure that the device is in "allocated" ... 13:12:35 this will be harder to fix 13:12:57 agreed, but will have time to look at it later this week 13:13:14 ok I will try to make again the test and test all the current patches 13:13:41 I assume there will be still race condition we need to handle 13:13:50 yes 13:14:10 we should at least target to have revert functionnality included in newton 13:14:17 agree 13:14:35 we can fix the racy conditions in next release 13:14:45 yeap 13:15:00 we also have this bug https://bugs.launchpad.net/neutron/+bug/1614092 in neutron 13:15:00 Launchpad bug 1614092 in OpenStack Compute (nova) "SRIOV - PF / VM that assign to PF does not get vlan tag " [Low,Confirmed] 13:15:22 and this is edand fix https://review.openstack.org/#/c/357244/ 13:15:32 please review it 13:15:49 will do 13:16:32 sorry it this bug https://bugs.launchpad.net/neutron/+bug/1614086 13:16:32 Launchpad bug 1614086 in neutron "SRIOV- when numvfs=0 SRIOV agent is failed to start " [Low,In progress] - Assigned to edan david (edand) 13:17:02 #topic Open Bugs 13:17:41 so we have some new bugs in neutron/nova 13:18:00 https://bugs.launchpad.net/nova/+bug/1606747 13:18:00 Launchpad bug 1606747 in OpenStack Compute (nova) "[sriov] nova does't regard second nic after booting" [Undecided,In progress] - Assigned to xhzhf (guoyongxhzhf) 13:18:10 https://bugs.launchpad.net/nova/+bug/1582822 13:18:10 Launchpad bug 1582822 in OpenStack Compute (nova) "PCI-PT : SRIOV enabled interface dev_name in pci whitelist does not give the product_id of PF for the direct-physical_network , it always take the VF's product_id" [Low,In progress] - Assigned to Chen Fan (fan-chen) 13:18:23 https://bugs.launchpad.net/nova/+bug/1572826 13:18:23 Launchpad bug 1572826 in OpenStack Compute (nova) "dev name in the pci whitelist is not honored for SRIOV" [Undecided,Incomplete] - Assigned to Moshe Levi (moshele) 13:18:34 https://bugs.launchpad.net/neutron/+bug/1614092 13:18:34 Launchpad bug 1614092 in OpenStack Compute (nova) "SRIOV - PF / VM that assign to PF does not get vlan tag " [Low,Confirmed] 13:18:58 so most of them are low priority 13:19:12 * sfinucan apologises for being extremely late, but has read the logs 13:19:35 hi sfinucan 13:19:39 hi 13:20:00 lbeliveau: I want to talk about this one https://bugs.launchpad.net/neutron/+bug/1614092 13:20:00 Launchpad bug 1614092 in OpenStack Compute (nova) "SRIOV - PF / VM that assign to PF does not get vlan tag " [Low,Confirmed] 13:20:33 so if you do physical-direct you need to set the vlan on the guest 13:20:43 yeah, not sure if it is a bug, in case a PF passthrough, the guest has to set the vlan 13:21:13 the only thing that could be automated (I think) is to use cloudinit for setting it up 13:21:15 do you know of a better solution or just to document this 13:21:25 I'll look into it 13:21:34 that what I was thinking 13:22:13 and I wanted to move this to documentation 13:22:28 agreed, that makes sense to document this 13:22:38 this also will complicate testing in CI 13:23:08 mmmm right, we'll need to find a way to automate this, let me think about it 13:23:25 yes I will do it too :) 13:24:08 also there is a bug to improve the NUMA scheduling https://bugs.launchpad.net/nova/+bug/1614882 13:24:08 Launchpad bug 1614882 in OpenStack Compute (nova) "NUMA node scheduling problem" [Undecided,New] 13:25:35 sfinucan, snikitin: are you planing to work on this maybe in the next cycle ? 13:25:40 yeah, we've been discussing that internally 13:25:48 looks like a feature for next cycle, yes 13:25:56 i.e. it's a feature request, not a bug 13:26:17 sure 13:26:27 that was my understanding, so when you have a spec ready please add us :) 13:26:38 FYI, we might need a blueprint for this 13:26:47 ok if you need help with spec 13:26:55 we have an in-house solution for that that is not upstreamed yet, can help 13:27:04 ok cool ;) 13:27:11 lbeliveau: Yeah, definitely a full blueprint/spec process 13:27:16 lbeliveau: very good to know :) 13:27:44 sfinican: we can co-author it if you want 13:27:56 #topic Specs for Review 13:28:39 I put a list of SR-IOV spec that are proposed in Ocata and was proposed before 13:28:57 see https://etherpad.openstack.org/p/sriov_meeting_agenda Under Specs for Review 13:30:31 I planning to focus on SR-IOV HA https://review.openstack.org/#/c/182242/ 13:30:55 and also the enhance PCI passthrough whitelist to support regex https://review.openstack.org/#/c/350211/ 13:31:36 if you know/have SR-IOV spec that are missing please add them to https://etherpad.openstack.org/p/sriov_meeting_agenda 13:31:57 so we can keep track of things 13:32:12 anyone have something to add here 13:32:17 I'm good 13:32:21 ditto 13:32:42 ok cool :) 13:32:51 #endmeeting