| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-cyborg - 2017-05-17 | |||
| 16:28:28 | openstack | Minutes: http://eavesdrop.openstack.org/meetings/openstack_cyborg/2017/openstack_cyborg.2017-05-17-15.03.html | |
| 16:28:30 | openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/openstack_cyborg/2017/openstack_cyborg.2017-05-17-15.03.txt | |
| 16:28:32 | openstack | Log: http://eavesdrop.openstack.org/meetings/openstack_cyborg/2017/openstack_cyborg.2017-05-17-15.03.log.html | |
| #openstack-cyborg - 2017-05-22 | |||
| 16:25:25 | crushil | jkilpatr, zhipengh[m] Do you care what hardware I use to write the generic driver? | |
| 22:52:43 | ttk2[m] | crushil: shouldn't it be hardware agnostic? | |
| 22:54:31 | zhipengh[m] | The generic driver should provide a set of basic functionality that are common among accelerators | |
| 22:54:52 | zhipengh[m] | It should be hardware agnostic | |
| 22:55:03 | ttk2[m] | IMO the generic driver should just be stubs for attach/detach install/uninstall | |
| 22:55:12 | zhipengh[m] | Ya basically | |
| #openstack-cyborg - 2017-05-23 | |||
| 15:19:22 | jkilpatr | -2 from a linters run? wow | |
| 15:19:28 | jkilpatr | most stick to just -1 | |
| 15:21:42 | zhipengh[m] | CI has been acting crazy.. | |
| #openstack-cyborg - 2017-05-24 | |||
| 15:00:07 | crushil | \o | |
| 15:00:34 | zhipeng_ | ho ho ho | |
| 15:00:45 | zhipeng_ | #startmeeting openstack-cyborg | |
| 15:00:46 | openstack | Meeting started Wed May 24 15:00:45 2017 UTC and is due to finish in 60 minutes. The chair is zhipeng_. Information about MeetBot at http://wiki.debian.org/MeetBot. | |
| 15:00:47 | openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | |
| 15:00:49 | openstack | The meeting name has been set to 'openstack_cyborg' | |
| 15:00:52 | jkilpatr | morning everyone | |
| 15:00:56 | zhipeng_ | #topic Roll Call | |
| 15:01:06 | zhipeng_ | morning , evenning and afternoon :) | |
| 15:01:11 | zhipeng_ | #info Howard | |
| 15:03:43 | jkilpatr | #info Justin | |
| 15:04:01 | crushil | #info Rushil | |
| 15:05:23 | zhipeng_ | anyone else ? | |
| 15:07:18 | zhipeng_ | okey then let's proceed | |
| 15:07:40 | zhipeng_ | we have 2 hours long meeting usually anyway :P | |
| 15:07:51 | zhipeng_ | #topic BP Discussion | |
| 15:08:00 | jkilpatr | we should just have a 24 hour meeting and call it an IRC channel | |
| 15:08:06 | zhipeng_ | hahaha | |
| 15:08:23 | jkilpatr | Anyways Rushil I think you need to wf 0 that commit then wf 1 it again to get the gates to rerun | |
| 15:08:25 | zhipeng_ | automatic ending every 24 hours and then start again | |
| 15:08:46 | zhipeng_ | i just did a recheck | |
| 15:08:56 | zhipeng_ | sometimes the gate is just behaving weird | |
| 15:09:07 | jkilpatr | if they still don't work I'll go and talk to the people in #openstack-infra they are always helpful | |
| 15:09:30 | jkilpatr | conductor and agent look good. | |
| 15:09:45 | jkilpatr | cyborg nova needs a meeting of its own in my opinion | |
| 15:09:56 | zhipeng_ | absolutely agree | |
| 15:10:28 | zhipeng_ | i just +2 the agent one, looking at the conductor spec | |
| 15:11:16 | jkilpatr | I think the Cyborg/Nova spec might be served by having us play around with writing some code and trying stuff. But that's a cart/horse issue. | |
| 15:11:23 | zhipeng_ | jkilpatr rushil I will nominate you guys as core reviewer as well today, so that we could move the development faster | |
| 15:11:53 | crushil | cool, thanks zhipeng_ | |
| 15:11:57 | zhipeng_ | jkilpatr you mean we do it in parallel ? | |
| 15:12:31 | crushil | Is there a deadline for the code to get in? End of Pike? | |
| 15:12:42 | jkilpatr | zhipeng_, kinda I mean we start writing stuff with merged specs and play around with the nova stuff, when we have a decent idea of the pattern we want to use we can formalzie it and merge it. | |
| 15:12:55 | jkilpatr | I think the issue is that there are enough moving parts that's hard to think about in a totally abstract sense. | |
| 15:13:09 | jkilpatr | at the same time it's not really the best practice to do things like that. But whatever gets the code working | |
| 15:13:10 | zhipeng_ | yes I agree | |
| 15:13:28 | zhipeng_ | otherwise we would just be talking in abstraction | |
| 15:13:50 | jkilpatr | which in principle is fine, but right now we're talking in abstract circles on this spec.... | |
| 15:14:38 | zhipeng_ | ah it seems I don't have the gerritbots here ..., but a quick update, jenkins shines upon the agent spec just now :P | |
| 15:14:51 | zhipeng_ | jkilpatr agree | |
| 15:17:13 | jkilpatr | crushil, you started work on the driver? | |
| 15:18:00 | crushil | jkilpatr, I am still deciding what hardware backend I should use as the prototype to start development | |
| 15:18:09 | crushil | Any suggestions? | |
| 15:18:45 | crushil | I asked this on the channel yesterday as well | |
| 15:18:53 | jkilpatr | ok this is a good subject to discuss | |
| 15:18:56 | zhipeng_ | have you looked into virtio ? | |
| 15:19:18 | jkilpatr | how generic should the generic driver be? | |
| 15:19:48 | jkilpatr | maybe we need a stub driver that's literally nothing but stubs for all the required functions then a generic driver that's a very basic driver implementation for $something | |
| 15:20:39 | crushil | Hmmm, would the "stub" driver require another spec? | |
| 15:20:48 | jkilpatr | nah it's just the current spec | |
| 15:21:12 | jkilpatr | I mean the current spec specifies driver concepts, the stub driver is just the simplest thing that implements those concepts but does nothing at all useful in life | |
| 15:21:23 | jkilpatr | I would think a ' | |
| 15:21:26 | jkilpatr | sorry | |
| 15:21:34 | jkilpatr | a 'generic' driver would be pci attachment though right? | |
| 15:22:05 | crushil | That's my understanding | |
| 15:22:18 | jkilpatr | ok I see how virtio fits into that picture, just took some googlign. | |
| 15:22:49 | jkilpatr | sounds good, stub is lower priority (not that it should be any effort really) | |
| 15:23:21 | crushil | What's the deadline for this development work though? End of Pike? | |
| 15:23:51 | jkilpatr | specs -> pike ; basic functionality -> qeens from what I understand | |
| 15:24:17 | zhipeng_ | i was thinking basic functionality at the end of pike | |
| 15:24:21 | zhipeng_ | at least we should try | |
| 15:24:46 | jkilpatr | isn't the pike deadline in like a month? or am I off | |
| 15:25:05 | jkilpatr | I guess that's possible | |
| 15:26:27 | zhipeng_ | no that was the spec deadline | |
| 15:26:47 | zhipeng_ | the end of cycle would be around Sep ? | |
| 15:26:49 | crushil | Hmmm, that would be complicated. I will be on PTO between Mid June to Mid July. Gone for almost a month | |
| 15:27:16 | crushil | August 28 is end of Pike I believe | |
| 15:27:28 | zhipeng_ | anyways let's try to shoot for Sep finish | |
| 15:27:30 | jkilpatr | oh man we have more time than I thought | |
| 15:27:35 | jkilpatr | that's doable | |
| 15:27:37 | zhipeng_ | Sep 11th is the PTG | |
| 15:27:39 | zhipeng_ | ya | |
| 15:27:44 | jkilpatr | where is it this time? | |
| 15:27:49 | zhipeng_ | lucky we are not doing actual releases :P | |
| 15:28:00 | crushil | Denver | |
| 15:28:41 | jkilpatr | can't say for sure if I'll be there yet. | |
| 15:28:59 | crushil | I'm still unclear on this, which backend should I use for the driver? VirtIO or a specific hardware? | |
| 15:29:23 | jkilpatr | well it's supposed to be generic, use virtIO | |
| 15:29:43 | crushil | ok | |
| 15:30:07 | zhipeng_ | i think virtio would be preferred choice now | |
| 15:34:04 | zhipeng_ | crushil you might consider what justin suggested for the generic patch | |
| 15:34:14 | zhipeng_ | gate god did not shine upon you today :P | |
| 15:34:19 | crushil | Lol | |
| 15:34:28 | zhipeng_ | two of J's specs are both merged | |
| 15:36:03 | jkilpatr | \0/ | |
| 15:36:40 | jkilpatr | I have experience pleasing Zull the linter god. | |
| 15:36:44 | jkilpatr | Zuul* | |
| 15:37:27 | crushil | It's unable to install dependencies for some reason | |