13:03:12 <baoli> #startmeeting PCI passthrough 13:03:12 <openstack> Meeting started Tue Aug 26 13:03:12 2014 UTC and is due to finish in 60 minutes. The chair is baoli. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:03:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:03:15 <openstack> The meeting name has been set to 'pci_passthrough' 13:03:21 <baoli> Hi, everyone 13:03:28 <irenab> hi 13:03:51 <irenab> baoli: How is going? 13:04:05 <baoli> irenab, let me report my status. 13:04:28 <irenab> baoli: I meant generally, not patches :-) 13:04:55 <baoli> irenab, I don't understand 13:05:39 <baoli> irenab, sorry, my bad 13:06:06 <irenab> baoli: never mind 13:07:00 <baoli> yea, the summer is about to be over, and the school starts tomorrow. 13:07:56 <irenab> baoli: cool, we have few more days till back to school 13:09:03 <baoli> irenab, that's better. My son is not excited about going back to school 13:09:23 <irenab> baoli: :-) 13:09:29 <irenab> smart guy 13:09:34 <heyongli> hi 13:09:43 <baoli> heyongli, hi 13:09:57 <baoli> ok, I think we can get started 13:10:03 <irenab> +1 13:10:34 <baoli> status wise, you probably have seen the activities from the review. 13:11:00 <baoli> it's getting closer, but a couple of sticking issues 13:11:50 <baoli> The first one is about the use of the system metadata for pci requests which is a 255 bytes json doc. And the concern is the out of bound issue 13:12:12 <irenab> baoli: I have some concern regarding the patch: https://review.openstack.org/#/c/107466/43/nova/virt/libvirt/vif.py 13:12:41 <baoli> The second one is about the data structure of the requested network, adding the request_id into the existing tuple 13:13:11 <baoli> irenab, what about it? 13:13:34 <irenab> baoli: can you please take a look at Itzik comment on patch#42 13:13:54 <heyongli> baoli, i will recheck the 2 concerns to find out what we can do 13:14:14 <irenab> after rebase you missed the cleanup for plug_hw_veb and unplug_hw_veb 13:14:43 <irenab> as code is now, it fails for hw_veb vif_type 13:15:40 <baoli> irenab, hmm, I changed it, I need to find out why the change was not uploaded. 13:15:59 <irenab> baoli: thanks! 13:18:13 <baoli> heyongli, I'm hoping that Dan can agree with addressing them separately. 13:18:55 <heyongli> oh, stick one, we can negotiate and same time try find a solution or better implementation, anyway , we try 13:20:05 <irenab> AFAIK September 4 is last day for code to merge 13:21:01 <heyongli> very tight schedule for us 13:21:08 <irenab> But bug fixes can land later 13:22:09 <baoli> we have only a few days left. 13:22:38 <irenab> I think probably better to fill bugs for the issues to resolve them later, if it is not blocking the solution of course 13:22:59 <heyongli> worth to try 13:22:59 <baoli> irenab, I'm hoping that the cores think the same way 13:23:41 <baoli> I will have several bugs/enhancements to file, for sure 13:24:10 <baoli> any thing else that you guys want to talk about? 13:24:28 <irenab> there were few action items we took last week 13:24:55 <baoli> ireanb, can you recap? I didn't look at last week's log yet. 13:25:13 <irenab> I need to start the list of next work items in the SR-IOV networking area 13:25:36 <baoli> irenab, on the neutron side? 13:26:00 <irenab> rpothier: to investigate on HEAT to configure VM with port vnic_type = direct 13:26:18 <irenab> baoli: no, as a further functionality as well 13:26:43 <irenab> once basic support including nova patches you push will be merged, what is next 13:27:14 <baoli> irenab, we have a list on the meeting wiki, but may not be complete. We also talked about rewriting the ml2 driver 13:27:17 <irenab> I hope to cover it during this week, and send an email for you guys to add 13:27:55 <irenab> baoli: refactoring of ML2 plugins defenitely in the list 13:28:07 <baoli> irenab, cool 13:28:26 <irenab> baoli: I'll start from the wiki 13:28:42 <baoli> irenab, send the list to the ML, and we can start from there 13:28:56 <irenab> baoli: sure 13:29:23 <rpothier> irenab: I looked into Heat, vnic_type can be configured currently, however a OS::Neutron::pci-port type would be better 13:30:07 <irenab> rpothier: can you please explain or maybe share some HEAT template to define VM with SR-IOV port? 13:30:46 <irenab> rpothier: Does it require to add support to HEAT? 13:31:55 <rpothier> I have a basic template at http://paste.openstack.org/show/100473/ 13:33:17 <irenab> rpothier: nice! 13:33:20 <heyongli> thanks 13:33:46 <irenab> rpothier: Do you suggest to add OS::Neutron::pci-port ? 13:34:02 <rpothier> irenab: yes 13:34:56 <rpothier> it's much clearer than adding parameters to value_spec 13:35:19 <irenab> rpothier: so the Template can be simplified, right? 13:35:29 <rpothier> irenab: yes 13:35:32 <heyongli> sriov is tech detail , pci thing might close to user, seems good 13:36:32 <irenab> rpothier: Is it something you want me to put on the list of next tasks, or plan to work on it during Juno? 13:37:40 <irenab> as long as there is a way to define it, I do not think its urgent to add 13:37:49 <rpothier> irenab: yes, you can put it on the list, not sure on timing 13:38:12 <irenab> rpothier: sure. Thanks a lot for checking 13:38:59 <irenab> do you know of some task that is critical to complete for our current support? 13:41:45 <irenab> Let's hope that nova patches will be merged soon and we can start focusing on next items 13:42:22 <baoli> let's keep our finger acrossed 13:42:35 <baoli> s/acrossed/crossed 13:43:01 <baoli> and that's the highest priority 13:43:05 <baoli> anything else? 13:43:29 <irenab> not for me 13:43:33 <heyongli> no for me 13:43:40 <baoli> good day and good night. See you guys next time 13:43:48 <irenab> see you 13:43:54 <baoli> #endmeeting