13:01:18 <moshele> #startmeeting sriov 13:01:18 <openstack> Meeting started Tue Jun 7 13:01:18 2016 UTC and is due to finish in 60 minutes. The chair is moshele. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:19 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:01:21 <openstack> The meeting name has been set to 'sriov' 13:01:35 <moshele> hi everyone 13:01:39 <gjayavelu> hi 13:01:56 <smatov> hello 13:02:05 <saneax> Hi Everyone 13:02:16 <moshele> #topic agenda Improving SR-IOV/PCI Passthrough CI 13:03:04 <moshele> so update on the Mellanox SR-IOV CI we have problem to move to containers we are still debugging the libvirt issue 13:03:28 <moshele> is any one from Intel CI here? 13:03:50 <moshele> is sean-k-mooney, sfinucan here ? 13:03:57 <moshele> I guess not :( 13:04:45 <moshele> so we are currently missing multi node CI for SR-IOV :( 13:05:09 <moshele> anyone have update on this topic? 13:05:32 <lbeliveau> I wish I had, but not at the moment 13:05:59 <moshele> #topic bug fixes 13:06:22 <moshele> so the pci resize patch was merged last week :) 13:06:46 <moshele> lbeliveau: is working on the pci migration patches 13:06:48 <lbeliveau> ye ! 13:07:29 <moshele> lbeliveau: I review them yesterday had some questions on them 13:07:56 <lbeliveau> moshele: I'm taking care of that atm, will post new patch set later today 13:08:32 <moshele> ok cool, I didn't test the pci migration, only with direct port at the moment 13:09:16 <moshele> I hope to continue this week 13:09:32 <lbeliveau> anybody else has patches that need reviews ? 13:10:26 <saneax> hello from me, we are working towards supporting sriov in tripleo 13:11:07 <moshele> we have also Numa patch Evaluate 'task_state' in resource (de)allocation - https://review.openstack.org/#/c/323269/ 13:11:52 <saneax> this the spec file we are working on for tripleo, would request some comments 13:11:54 <saneax> https://review.openstack.org/#/c/313872 13:12:02 <moshele> saneax: let talk on the next topic this is bug fixes ok 13:12:14 <moshele> anyone else on bug fixes ? 13:12:20 <saneax> ok moshele 13:12:51 <moshele> ok let move one to spec 13:13:09 <moshele> #topic Specs for review: 13:13:21 <moshele> saneax: ok your up :) 13:13:32 <saneax> thanks moshele 13:13:51 <saneax> We are trying to get SRIOV functionality within the tripleo 13:14:04 <saneax> our spec file is https://review.openstack.org/#/c/313872 and we are looking for comments in it 13:14:51 <lbeliveau> saneax: i'll have a look 13:14:59 <saneax> thanks 13:15:12 <moshele> saneax: so this is just to configure the compute to support SR-IOV? 13:15:25 <saneax> yes moshele 13:15:57 <moshele> saneax: so basically it related to ironic right? 13:16:13 <saneax> so, it looks at ironic, THT, puppet 13:16:59 <moshele> can you explain how is the flow with tripleo ? 13:17:18 <saneax> overall a operator can configure SRIOV functionality on the cluster is the end goal. We would like to support mixed cluster, with sriov and without sriov computes 13:17:43 <saneax> so, at present moment introspection bits are not utilized in THT 13:18:11 <saneax> our idea is to to just give heads up to operator on the PCI details on the introspection 13:18:32 <moshele> how do you inspect the nodes? with ironic-inspector? 13:19:07 <moshele> so it is manually? 13:19:20 <moshele> nevermind I will review the spec and comment 13:19:29 <saneax> yes, its manual 13:20:21 <saneax> The idea is to add most of the manual steps in the installer 13:20:34 <saneax> which is now a post install after overcloud deploy 13:20:44 <moshele> and what will configure the driver and set SR-IOV VF - heat? 13:21:07 <saneax> so so as a first cut we would have all the drivers in the overcloud image 13:21:15 <saneax> we would do this for RH flavors 13:21:48 <moshele> but you need to set the number of VF in each compute 13:22:20 <saneax> yes, so that we are taking default all the pci addresses that is being exposed 13:22:48 <moshele> I guess I need to read the spec :) 13:22:53 <saneax> :) 13:22:58 <moshele> anything else on the matter? 13:23:09 <saneax> thats all, thanks 13:24:07 <moshele> ok, so I am still working on the User-controlled SR-IOV ports allocation spec I didn't upload any patch yet. I hope to do it next week 13:24:30 <moshele> anyone else has specs that he want to raise here? 13:24:49 <gjayavelu> moshele: i'll be moving vmware sriov spec to ocata 13:25:13 <moshele> gjayavelu: did you talked about it in the nova meeting? 13:26:10 <moshele> so for nova it is freeze for spec in newton as explained in the ML 13:26:29 <gjayavelu> moshele: No, since i couldn't make it to last week's schedule. 13:26:54 <gjayavelu> i tried pinging jay, no luck 13:26:59 <moshele> gjayavelu: I would suggest to start POC it 13:27:10 <gjayavelu> sure, i'm working on that 13:27:18 <moshele> cool :) 13:27:40 <moshele> any other specs to address ? 13:28:49 <moshele> saneax: I added you spec to the agenda for next meeting 13:29:09 <moshele> anything else? 13:29:51 <saneax> thanks 13:30:10 <moshele> ok I think we are done for today :) 13:30:21 <moshele> #endmeeting