15:00:17 #startmeeting gantt 15:00:18 Meeting started Tue Apr 22 15:00:17 2014 UTC and is due to finish in 60 minutes. The chair is bauzas. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:19 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:21 The meeting name has been set to 'gantt' 15:00:26 hi all 15:00:35 hi 15:00:36 hi 15:00:39 hi 15:00:57 o/ 15:01:07 nice to see people there :) 15:01:27 waiting one more min and then we start 15:01:29 finally worked out the time zone 15:01:35 bauzas: Hi :) 15:01:49 YorikSar: hi ! j 15:02:29 #topicĀ Open action items 15:02:56 #link http://eavesdrop.openstack.org/meetings/gantt/2014/gantt.2014-04-15-15.00.html 15:03:16 quite simple, there were 3 actions from one single person last week :) 15:03:32 luckily, I know him 15:04:03 so, all, you can look at https://wiki.openstack.org/wiki/Gantt 15:04:16 #link https://wiki.openstack.org/wiki/Gantt#Developer_Summit_Sessions 15:04:29 here is the etherpad link for proposals 15:04:44 I went thru all the open subjects 15:04:49 maybe I missed some 15:05:04 so feel free to add/amend the etherpad if so 15:06:01 the best would be that mikal and johnthetubaguy would take note of this page 15:06:04 Not sure if http://summit.openstack.org/cfp/details/262 can be put to gantt? 15:06:28 jay-lau-513: good question 15:06:32 http://summit.openstack.org/cfp/details/262 is for dynamic scheduling 15:06:58 jay-lau-513: I think that some gantt people would love joining the discussion 15:07:11 jay-lau-513: even if it's not purely related to gantt 15:07:35 Great, the dynamic scheduler can work for nova, cinder even neutron etc 15:07:38 jay-lau-513: IMHO, this service should be linked someway to gantt 15:07:41 262 is related to 400 15:07:45 but we can first start from nova for this 15:07:54 I see Gantt as preparing for both 15:08:03 mspreitz: +1 15:08:26 the problem is see here is the chicken-and-egg one 15:08:44 as gantt is currently not yet delivered, it can't be leveraged for this purpose 15:09:01 but possibly it should help 15:09:38 I do not see a circle of dependency 15:09:46 262 and 400 depend on Gantt, not the other way around 15:10:34 well, let's discuss about this topic later in the meeting 15:10:51 mspreitz yes, 262 is for dynamic schdulig and seems 400 is still for holistic static placement 15:11:08 I also had an action for linking Gantt wiki with Nova one 15:11:19 that's done : 15:11:33 #link https://wiki.openstack.org/wiki/Nova#Active_Sub-teams: 15:12:14 the last action I had, about creating a 2nd blueprint for the sched forklif 15:12:34 #link https://blueprints.launchpad.net/nova/+spec/isolate-scheduler-db 15:12:47 the nova-specs patch has to be sent 15:13:31 let's move quickly to the next topic 15:13:42 so, we'll have time for discussions :) 15:13:55 #topic Status on forklift efforts 15:14:18 #link https://review.openstack.org/#/c/82133/ 15:14:42 if people have time for reviewing it, that would be great 15:15:18 I made a dependency on https://review.openstack.org/#/c/86988/4 15:15:43 any comments yet ? 15:16:10 I take it for a no :) 15:16:30 as said, I also began working on another blueprint 15:16:44 https://blueprints.launchpad.net/nova/+spec/isolate-scheduler-db 15:17:11 the idea is to have the scheduler only accessing compute_nodes db table 15:17:28 bauzas: I guess that should be related to no-db-scheduler 15:17:47 YorikSar: that's a separate blueprint 15:17:51 IMHO 15:18:28 YorikSar: at the moment, scheduler is not only accessing compute_nodes, but also aggregates, services etc. 15:18:40 bauzas: Sure. I see. 15:18:52 YorikSar: so there is need to modify the calls 15:19:34 YorikSar: for baby-stepping, I was planning to only refactor the changes and isolate all the calls into the next scheduler client lib 15:19:48 YorikSar: but that needs to be debated 15:19:55 both in terms what and how :) 15:20:39 YorikSar: correct me if I'm wrong, but no-db sched is planning to rewrite accesses to compute_nodes table ? 15:20:47 YorikSar: only 15:21:17 bauzas: Yes. It isolates only nodes/services/hoststates. 15:21:37 YorikSar: I saw your patches 15:21:55 YorikSar: I would love discussing it in the next second topic 15:22:25 YorikSar: because we need to see when you plan to deliver it 15:22:29 bauzas: Looking forward to it :) 15:23:08 ok, any other comments about the forklift ? 15:23:57 fyi, as the sched-lib blueprint is close to be accepted, I'll publish a first patch for implementing it, next week 15:24:15 ok, next topic 15:24:25 #topic Juno summit design sessions 15:24:31 again, the etherpad 15:24:39 #link https://etherpad.openstack.org/p/Gantt-summit-sessions 15:25:18 so, jay-lau-513, do you prefer I remove mention to 262 ? 15:25:53 bauzas, you mean remove 262 from gantt session? 15:26:05 jay-lau-513: from the etherpad, at least :) 15:26:24 we still have to discuss with mikal to see how many sessions we will have 15:26:35 and if there is need to group some proposals 15:27:00 bauzas, ok, but does gantt will handle dynamic schduling in the future? 15:27:12 maybe russellb is knowing how many sessions we will have for scheduler ? 15:27:53 jay-lau-513: maybe that could be discussed during http://summit.openstack.org/cfp/details/140 15:28:08 jay-lau-513: I mean, Gantt scope should be considered therer 15:28:10 there 15:28:22 bauzas: OK, no problem 15:28:52 jay-lau-513: so, do we keep track of your proposal as gantt related subject ? 15:29:11 I think that we can merge it to 140 15:29:21 bauzas: make sense? 15:29:52 jay-lau-513: when do you plan to implement 262? 15:30:17 My estimation is that scheduling will get few slots, 142, 2562, 400 will get merged into 140 15:30:17 bauzas: You mean implement code or just spec :-) 15:30:34 jay-lau-513: I mean implementation 15:30:34 s/2562/262/ 15:30:39 mspreitz: Agree 15:31:02 bauzas: In Juno if approved, but not sure in which sprint. 15:31:17 jay-lau-513: that sounds a big baby here :) 15:31:29 bauzas: I also discussed with someone, many of them prefer a new project for this :) 15:31:43 jay-lau-513: who's someone ? :D 15:31:58 bauzas: let me check my email 15:31:58 jay-lau-513: ok, I see 15:32:04 jay-lau-513: no worries 15:32:18 jay-lau-513: I remember the discussion in the ML 15:32:38 jay-lau-513: but as mspreitz said, it could possibly need Gantt 15:32:46 for implementation 15:33:25 jay-lau-513: ok, let's amend the etherpad for merging 140 and 262, and let's discuss it in the etherpad 15:33:28 bauzas: Exactly, gantt is th ebase 15:33:31 the base 15:33:42 bauzas: OJK 15:33:45 ok 15:34:00 bauzas: http://markmail.org/message/5zotly4qktaf34ei 15:34:14 #action bauzas to merge http://summit.openstack.org/cfp/details/262 into http://summit.openstack.org/cfp/details/140 in the etherpad 15:34:54 * bauzas loves to see his name 15:35:12 ok, any other sessions to mention? 15:35:15 I have been thinking about 99 15:35:25 that email just cited mixes 99 and 262 15:35:36 I was willing to accept 99 as a separete, small scope thing 15:35:42 but that email links to bigger ideas 15:35:54 sounds lik 15:36:03 I agree that 99 can be seen as part of bigger things 15:36:04 mspreitz: sounds like 99 is the first step for 262 15:36:13 Exactly, let's do 99 as a baby step first 15:36:43 will also amend the etherpad if so 15:36:44 bauzas mspreitz yes, 99 is the first step for many projects ;) 15:37:53 well, the no-db sched effort would love discussing about 99, I bet :) 15:38:02 maybe I'm wrong, YorikSar ? 15:38:03 :) 15:38:12 http://summit.openstack.org/cfp/details/99 15:38:47 bauzas: I won't come to summit but someone will definitelly come to discuss further directions in that. 15:38:47 the nova spec for 99 is here https://review.openstack.org/#/c/88983/ 15:39:27 YorikSar: well, the idea of persisting hints would impact no-db sched efforts I suppose 15:39:43 so, indeed, that would be worth discussing it at summit time 15:40:24 jay-lau-513: I added myself to the reviewers for the spec 15:40:30 bauzas: I'm not sure about that. no-db part is about host states, not about resources themselves. 15:40:32 bauzas: yes, I think 99 does deserve a small session though it is straigforward, but it has some interation with other blueprints 15:40:43 bauzas: great :) 15:40:48 jay-lau-513: will try to take time to review it 15:41:03 anyway, that's a good transition for... 15:41:08 so many Jays... my IRC is just lighting up. :) 15:41:15 #topic no-db scheduler 15:41:36 YorikSar: you're up 15:41:37 :) 15:41:49 jaypipes sorry for the confuse :-) 15:41:54 YorikSar: glad to see you took the blueprint 15:41:55 I've picked up Alexey's work in Nova. 15:42:16 Currently it looks like technical part works. 15:42:26 YorikSar: could you please summarize a bit on what you're doing? 15:42:32 (at least Jenkins is not complaining) 15:42:37 YorikSar: I saw some patches 15:42:37 jay-lau-513: should that change be linked to bp? 15:43:02 YorikSar: do you need reviewing help ? 15:43:04 mspreitz which change? 15:43:07 https://review.openstack.org/#/q/topic:bp/no-db-scheduler,n,z - this patch series 15:43:27 I'd appreciate anyone besides bots to take a look at it. 15:43:51 #link https://review.openstack.org/#/q/topic:bp/no-db-scheduler,n,z 15:43:59 (for minutes) 15:44:02 I still have 2 -2's because I need to write blueprint to nova-specs and get it approved (again) 15:44:26 YorikSar I will add myself as reviewer 15:44:29 bauzas: thanks. I forget about those all the time 15:44:31 YorikSar: ok, let us know when you're done with the spec, so we can review it 15:44:53 YorikSar: that's already in my review list 15:45:05 I asked our Rally team to stress-test that code to see if there're more errors to be fixed. 15:45:22 YorikSar: cool 15:45:39 jay-lau-513, bauzas: Thanks, I'm looking forward to your -1's (with comments) :) 15:45:55 YorikSar: you know how I love reviewing your code ;) 15:46:18 bauzas: It's Alexey's code with couple of bugs fixed by me 15:46:43 bauzas: So you'll have to try to love reviewing his code too ;) 15:46:50 YorikSar: ;) 15:47:14 I guess we can add an action item for me to publish spec. 15:47:49 #action YorikSar to publish nova-spec for no-db scheduling blueprint 15:47:57 there it is 15:48:03 bauzas: cool 15:48:11 thanks for your support :) 15:48:41 YorikSar: is that basically ready for reviewing or do you plan further bugfixing ? 15:49:06 YorikSar: as you mentioned, Jenkins is happy but still :) 15:49:31 bauzas: It's ready. Except for Memcached part that laks both unittests and real-world testing. 15:49:42 I'll mark that one as WIP 15:50:07 YorikSar: cool thanks 15:50:14 Oh, I can't :( 15:50:28 YorikSar: ok, no worries, just put a comment then 15:50:33 I'll -1 it :) 15:50:39 sure thanks 15:50:48 can we move to open discussion ? 15:50:53 sure 15:50:56 #topic opens 15:51:09 there, 10 mins to go 15:51:17 any subject to raise? 15:51:21 any miss I made ? 15:51:34 any complain to do ? 15:52:01 How often are we going to have these meetings? 15:52:42 Does the meeing time always UTC 15:00 every Tuesday? 15:52:43 every Tuesday 15:52:49 jay-lau-513: indeed 15:53:15 bauzas: ok 15:53:37 https://wiki.openstack.org/wiki/Meetings/Scheduler 15:53:39 Great. It's hard to remember when to expect every-other-week meetings. Good to have a weekly one :) 15:54:12 well, I think we'll only have one left before summit 15:54:40 maybe on May 6th, but that's close to summit dates 15:55:25 ok, any other things to discuss ? 15:55:54 PaulMurray: I had no time for reviewing yet (Monday was off in France), but I'll do it for your patches on extended RT 15:56:16 PaulMurray: I saw some new patchsets 15:57:01 PaulMurray: s/extended/extensible/ 15:57:49 ok, thanks all for your presence 15:58:08 let's discuss next week 15:58:16 thanks 15:58:18 #endmeeting