15:04:18 #startmeeting openstack-cyborg 15:04:18 Meeting started Wed Jun 21 15:04:18 2017 UTC and is due to finish in 60 minutes. The chair is zhipeng. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:04:18 15:04:19 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:04:21 The meeting name has been set to 'openstack_cyborg' 15:04:25 hey zhuli welcome 15:04:57 okey on Mike's note, I want to share something in case you missed from our last super long meeting minutes 15:05:26 So I met with Ben Plaff from OVS, and did an OVS Orbit episode with him, on Cyborg and OVS-DPDK 15:05:47 I updated our master deck to add some materials related to ovs-dpdk 15:06:02 #link https://docs.google.com/presentation/d/1xXjqlMaXjXe7mn-iwoETGcTPnTbqCDfpaT6yqlZ8srE/edit?usp=sharing 15:06:03 cool, I'll have to go and look at that. 15:06:21 so the podcast will come out likely around early next month 15:06:47 and given Ben's status in open source networking, I think this is a good publicity for Cyborg :) 15:07:56 Agreed, hopefully we can have something to show at that time too. 15:08:40 yep :) 15:08:50 ok let's move on to our original agenda 15:08:59 #topic BP and coding discussion 15:09:04 #link https://review.openstack.org/#/q/status:open+project:openstack/cyborg 15:09:26 For the api/db spec 15:09:46 I will address and provide a update for crushil and zhuli's comments 15:09:51 hopefully tmr 15:10:05 any other major concerns over the api spec ? 15:10:37 I'm happy with it. 15:11:03 If we're down to correcting typos it means it's mostly done 15:11:09 cool 15:11:37 so on the stubs patches, any outstanding issue so far ? 15:12:18 I managed to patch my pep8 errors this morning, now I'm working on transport urls and how to configure them, it's more confusing than I expected configuration to be. 15:12:34 good news is once I get that working we can have a common configuration segment that takes care of it 15:13:25 great ! 15:13:42 #info agent stub https://review.openstack.org/473486 15:13:55 #info conductor stub https://review.openstack.org/472662 15:14:10 #info generic driver stub https://review.openstack.org/473186 15:15:42 crushil any problems from your side ? 15:15:46 regarding the api, how aligned is it with ETSI NFV specs? 15:15:56 Nope. I will fill out the driver as time goes by 15:15:58 should we consider going ahead and reviewing/landing the generic driver? I don't know what more we want from the dummy driver. 15:16:08 I've try to align with it as much as possible 15:16:34 jkilpatr I will take a look at that tmr my timezone 15:16:42 jkilpatr, We can, but we need agent and API code completely mapped out before the generic driver can be completely filled out 15:16:58 crushil actually not entirely 15:17:21 you could see my definitions in the api spec 15:17:59 anyways I will review and comment the patch see if we are good to go to land the first code 15:18:08 Ok, I will fill it out this week and remove the WIP 15:18:42 crushil, yes but adding onto it in the other patches is possible. I just don't want to end up with siloed commits where everyone is waiting on everything to work so they have to keep pulling stuff into their own commits 15:18:43 great 15:18:45 I was just waiting for the API and DB design proposal spec to merge before I filled it out 15:19:03 jkilpatr, Makes sense 15:19:06 lol I will accelerate my work on that 15:19:57 btw i saw in the news that NVIDIA release the PCIe version of their V100 15:20:24 That's great 15:20:29 this normal form factor means we could play with it, if there are vendors willing to provide it lol 15:20:49 oh another news 15:20:49 Btw I have a GPU sitting in the lab in one of my servers 15:20:56 or people willing to pay for it 15:20:58 ah cool crushil 15:21:02 It's a Tesla K80 15:21:12 I still need to get a gpu or two into my lab machines for testing 15:21:23 that is awesome enough for Cyborg :) 15:21:36 continue on my news reporting 15:21:50 I talked with OPNFV DPACC ptl the other day 15:22:06 LingLi? 15:22:10 yep 15:22:26 it turns out that it might be possible for NXP to donate some FPGA devices 15:22:27 Any plans to reanimate the project? 15:22:48 zhipeng, That's awesome news 15:22:49 so that we could start to do the Cyborg integration testing using OPNFV's lab facility 15:23:10 which is awesome and free for projects :) 15:23:20 On my end, I talked to Micron about adding a driver last week and talking to Xilinx today to have their driver in Cyborg 15:23:40 NokMikeR I'm trying to work with Lingli to restart DPACC on early July 15:23:52 crushil super cool man !! 15:23:56 sounds like there's lots of activity then. 15:24:06 If you guys want to join, I can send you the link to the meeting 15:24:21 crushil that'd be awesome 15:24:24 I might if I have time. 15:24:25 ok re DPACC 15:24:33 It might be too late for you zhipeng, but people in North America or Europe can join 15:24:45 crushil yes understood :) 15:25:09 and if we got meeting minutes that'd be perfect 15:25:40 NokMikeR your thought on DPACC ? 15:26:37 more complicated than originally thought - issue seems to be related to translating the gapi to something usefull. 15:27:02 okey that would be on the data plane side of things 15:27:10 in summary maybe a bit before its time? lots of missing parts etc. 15:27:13 I would leave that to you experts :P 15:27:18 lol 15:27:51 I think in jist Lingli said that as long as there are people willing to work on the project 15:28:05 she's more than willing to restart the project 15:28:26 so for gpai as long as there are folks actively working on it 15:28:31 Thats also important eg dont bother restarting unless theres really something to address. 15:28:34 maybe some of the problems could be fixed 15:28:37 what lead to it being stopped in the first place? 15:28:56 fdio mass migration? 15:28:58 people sorta drift away 15:29:05 ah maybe that 15:30:05 okey any other buisness 15:30:15 we could wrap up quick this time 15:30:41 oh on top of my head 15:30:50 the CFP for Sydney Summit is already open 15:31:00 and deadline is around July 15th I guess 15:31:16 let's all think of possible session topics 15:31:33 we could each summit one cyborg related talk 15:32:03 SO weird CFP is this early, but we gotta play by the rules 15:32:16 Or we can submit one joint talk where we have a basic POC by the time summit rolls around 15:32:27 that as well :) 15:33:49 okey if no other business I could close the meeting early today 15:33:58 nothing this side 15:34:19 leave a note here if you come up with anything, handles with [m] are always lurking here :) 15:34:59 thanks everyone 15:35:03 #endmeeting