03:03:22 <Li_Liu> #startmeeting openstack-cyborg
03:03:23 <openstack> Meeting started Wed Jan 23 03:03:22 2019 UTC and is due to finish in 60 minutes.  The chair is Li_Liu. Information about MeetBot at http://wiki.debian.org/MeetBot.
03:03:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
03:03:26 <openstack> The meeting name has been set to 'openstack_cyborg'
03:03:33 <Li_Liu> #topic Roll Call
03:03:40 <Li_Liu> #info Li_Liu
03:03:42 <Li_Liu> Hi Yumeng
03:03:44 <yumeng> #info Yumemg
03:05:03 <wangzhh> #info wangzhh
03:05:08 <xinranwang> #info xinranwang
03:05:33 <Li_Liu> wangzhh, do you know if coco is coming?
03:05:57 <wangzhh> Not sure, let me ping her.
03:07:45 <Li_Liu> #topic status of DB work
03:09:16 <Li_Liu> I think we need to merge this one https://review.openstack.org/#/c/625630/ once it's ready
03:09:45 <yumeng> Yep +1
03:09:53 <Li_Liu> I think after coco address wangzhh's comment, it can be merged
03:10:08 <wangzhh> Yup, I have a -1 on it.
03:10:51 <wangzhh> Sundar said that it's better to add UT...
03:10:54 <Li_Liu> and then we can merge these two: Add AttachHandle and ControlpathID objects	 + Add ExtArq and ARQ object
03:11:21 <xinranwang> yup, there is no other comment for coco's patch from my side
03:12:00 <wangzhh> My work depend on xinran and yumeng.
03:12:15 <yumeng> I will update mine today :)
03:12:22 <Li_Liu> based on https://storyboard.openstack.org/#!/story/2004248
03:12:40 <xinranwang> wangzhh:  i will address your comment today, and submit a new one
03:13:00 <Li_Liu> I think we still need Deployable, Device, and Accelerator
03:13:02 <wangzhh> Thx, xinran, yumeng
03:13:15 <Sundar> #info Sundar
03:13:19 <Sundar> Hi all
03:13:23 <wangzhh> Hi sundar.
03:13:29 <Li_Liu> HI Sundar
03:13:41 <Li_Liu> wait, yumeng, did you do the device already?
03:13:53 <Li_Liu> Device table I mean
03:13:55 <yumeng> Li_Liu: Device is included in my patch.
03:14:00 <Li_Liu> I see
03:14:07 <yumeng> Yeah already in my patch :)
03:14:25 <Li_Liu> make sure coco knows coz I see that table was assigned to her
03:14:45 <yumeng> Ops ok will check with her
03:14:46 <Sundar> yumeg: Did you leverage the pilot code?
03:15:06 <xinranwang> i suppose the table is assign to coco, and related object is what yumeng's doing
03:15:11 <Li_Liu> yumeng, and still need your Device Profile table
03:16:09 <Sundar> yumeng: Did you leverage the pilot branch code? The device profiles etc. are done there already
03:16:35 <Li_Liu> Sundar, I was trying to get it working, but still having some environment issue. will let you know if I can not resolve them
03:17:02 <yumeng> Sundar: yes I see it.
03:18:00 <Li_Liu> Hi shaohe_feng_
03:18:02 <yumeng> Sundar: and most are synced with yours
03:18:28 <shaohe_feng_> Li_Liu: evening
03:19:11 <Sundar> Cool. Li_Liu: Please LMK if you need any help. I tried it in 3 different devstack environments, 2 VMs and a physical host, before pushing the code
03:19:22 <Sundar> I know the environment can be tricky
03:19:27 <Li_Liu> sure thing
03:20:08 <Sundar> I need to drop off in 10 minutes today unfortunately. Please LMK what you need from me in this call
03:20:22 <zhipeng> Sundar is it your personal env for testing ?
03:20:32 <Sundar> zhipeng: Yes
03:20:47 <zhipeng> Neat
03:20:51 <Sundar> The physical host has an FPGA
03:21:22 <Li_Liu> Sundar, is that physical machine accessible from outside?
03:21:49 <Sundar> No :) It is inside an Intel lab, behind a firewall and a proxy
03:21:58 <shaohe_feng_> Sundar: can I access it?
03:22:32 <Sundar> Shaohe_feng_: Probably yes. Anybody within Intel network can reach it
03:23:04 <shaohe_feng_> Sundar: Good, can I have a try?
03:23:26 <Sundar> Sure. I will send you the IP address. I can also set you up as a separate user
03:24:02 <Li_Liu> Sundar, I might just bring up an aws vm and we can work together on it :P
03:24:14 <Sundar> For those without an FPGA, it is not too much of a hassle -- the VM will not come up because PCI passthrough fails, but you can see the debug output from libvirt driver with the domain XML
03:24:26 <zhipeng> Hybrid cloud
03:24:29 <Sundar> The domain XML should contain the hostdev device with the PCI address set up in Cyborg db
03:24:44 <openstackgerrit> Xinran WANG proposed openstack/cyborg master: Add AttachHandle and ControlpathID objects  https://review.openstack.org/630579
03:24:58 <Sundar> Li_Liu: Cool :)
03:25:29 <xinranwang> wangzhh:  just submit a new one, added the get_by_id function
03:25:58 <yumeng> Sundar: haha so cool
03:26:15 <wangzhh> Good, thanks xinranwang.
03:26:54 <Sundar> FYI, Nova developers are telling me that they don;t want Cyborg client as an intermediary either -- they must have misspoke in previous discussions. Instead, we wil provide code within Nova to call Cyborg API directly through a keystone adapter
03:27:33 <Li_Liu> it doesn't sound too bad tho
03:27:53 <Sundar> The API return value (JSON) will be converted by the 'accelerator' module within Nova. Similar to nova/network/neutronv2/api
03:28:12 <wangzhh> Sundar, does they have an agreement on it? :)
03:28:41 <Sundar> wangzhh: Easy to guess the answer :)
03:28:56 <Li_Liu> guess not ><
03:29:37 <Sundar> There is agreement that Nova code should not call Cyborg client. The differing views on whether Nova should call Cyborg API still needs discussion based on the pilot
03:30:09 <Sundar> Right now, we are getting reviews from a somewhat limited set of Nova developers, so we can clean up the smaller issues and take it to a broader audience
03:30:52 <Sundar> #link https://review.openstack.org/#/q/status:open+project:openstack/nova+bp/nova-cyborg-interaction
03:31:13 <zhipeng> So we strict client usage to admin ops ?
03:31:16 <wangzhh> Thx Sundar~
03:31:26 <zhipeng> Makes sense I guess
03:32:15 <Sundar> zhipeng: Yes. This is a better outcome.
03:33:51 <Li_Liu> Thanks a lot for the effort Sundar
03:34:10 <Sundar> Li_Liu: Welcome
03:34:11 <Li_Liu> I think we are at a much better position than last year Nov
03:34:55 <Sundar> I need to drop off now to attend to my family. Sorry about that.
03:35:08 <Li_Liu> np, go ahead man
03:35:13 <Sundar> Please feel free to ping me by email or Wechat if you have questions or comments
03:35:17 <zhipeng> Family first always
03:35:18 <Li_Liu> sure
03:35:26 <Sundar> Good day, y'all
03:35:32 <Li_Liu> :)
03:35:40 <Li_Liu> let's get back to the DB work status guys
03:36:01 <Li_Liu> we only miss couple tables now
03:36:13 <Li_Liu> from CoCo and yumeng
03:36:34 <xinranwang> which table?
03:36:49 <Li_Liu> Device Profiles
03:37:31 <xinranwang> Ah ok
03:37:35 <Li_Liu> and also refactor work on accelerators and deployable tables
03:38:18 <Li_Liu> let's tyr to get the patches merged by the end of this week
03:38:36 <Li_Liu> so that we can start the driver work
03:39:42 <Li_Liu> wangzhh, please ask if coco needs any help.
03:40:08 <Li_Liu> otherwise I can help some of her pieces
03:40:27 <zhipeng> Thumb up
03:40:27 <zhipeng> Have we cut the client and os-acc release yet ?
03:40:45 <Li_Liu> not yet
03:40:56 <Li_Liu> I will do that asap
03:41:27 <wangzhh> OK.
03:42:18 <zhipeng> Cool
03:42:26 <Li_Liu> I think that all we need to discuss for this meeting
03:42:33 <zhipeng> Fell free to ping Sean on wechat if you have any problems
03:42:42 <Li_Liu> #topic AOB
03:42:55 <Li_Liu> zhipeng, sure
03:43:14 <xinranwang> Btw, i see the deployable table is refactored, but not the object
03:43:39 <Li_Liu> I think coco is working on that
03:44:11 <xinranwang> ok cool
03:44:42 <Li_Liu> yumeng, just make sure you communicate with coco for the Device table you implemented.
03:45:04 <Li_Liu> That's it guys
03:45:25 <Li_Liu> have a good day/night where ever you are
03:45:26 <yumeng> Li_Liu: ok.
03:45:31 <Li_Liu> #endmeeting