09:00:17 #startmeeting blazar 09:00:18 Meeting started Tue May 8 09:00:17 2018 UTC and is due to finish in 60 minutes. The chair is masahito. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:00:19 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:00:21 The meeting name has been set to 'blazar' 09:00:26 #topic RollCall 09:01:36 o/ 09:01:42 masahito: hi 09:02:07 hiro-kobayashi: hi 09:02:13 hi all! 09:02:39 Today's agenda is 09:02:46 1. placement support 09:02:48 2. AOB 09:02:51 less topics 09:02:54 anything else? 09:03:01 bertys_, hiro-kobayashi: hello 09:04:05 #topic placement support 09:04:43 In last meeting hiro-kobayashi shared design idea for placement support. 09:05:00 https://etherpad.openstack.org/p/blazar-placement-design 09:05:35 and asked us to comment for the approach. 09:05:49 I commented this week. 09:06:10 hiro-kobayashi: any update for the design? 09:06:15 thanks for comments and adding details! 09:06:45 As we discussed before, I'd like to share the idea to openstack-dev ML before the Vancouver summit. 09:08:08 masashito: thanks. 09:08:22 My main question for the design is will host reservation ask user to choose private flavor instead of scheduler hint? 09:09:50 IIRC, scheduler filter is not recommended because it affects inter-operability. 09:10:23 So, I think it's better to use a private flavor. 09:10:51 However, it will increase the number of flavors... 09:12:18 yes. 09:13:15 Does 4 need a private flavor? 09:13:31 My concern is how host reservation create its flavors and count the usage. 09:13:54 How does the lease owner specify the allocation ID? 09:14:09 My understand is yes for instance reservation but no for host reservation. 09:15:13 IIRC, Nova doesn't have the feature now. 09:15:21 We need to propose it. 09:16:13 hiro-kobayashi, masahito: What about focusing first on instance reservation only? 09:17:31 bertys_: yes. 09:18:12 bertys_: make sense but it's better to consider both for basic design. 09:18:38 I also thought we need to handle 3 items with placement, freepool, host reservation and instance reservation, when I read the design. 09:20:21 masahito: Yes. And another item is capacity query. It's an advanced topic, thogh. 09:20:56 There could be another Nova-ish approach. Once we've written down our concern, it's good to share Nova team ASAP. 09:20:59 I mean, using placement for the 'reservation candidates' API 09:21:15 masahito: Agree. 09:22:46 hiro-kobayashi: do you have a planning date to share it to the team? 09:23:25 No, but before the summit. 09:25:30 If possible, I'd like to share it before next meeting because we can discuss the reaction from the team in weekly meeting and Blazar's forum session is on Monday in the summit. 09:25:56 s/in weekly meeting/ in the next weekly meeting/ 09:26:08 From the analysis, it seems right now that we have three options on the table for instance reservation: (#1 and #3) or (#2 and #3) or 4 as REQ3 is mandatory req IMHO 09:26:15 Do you mean Placement weekly meeting? 09:26:38 with (#1 and #3)= #5 09:27:40 sorry (#1 and #3) would be #6 ); 09:27:46 hiro-kobayashi: the blazar weekly meeting 09:28:53 bertys_: I thinks #1 can satisfy REQ 1-3. so #1, #4, #5 09:29:09 are the possible approach. 09:29:50 masahito: got it 09:30:22 #5 only works for the instance reservation, right? 09:31:19 masahito: I thinks it can be used for both of host and instance. However, host reservation has a problem of the number of flavor. 09:33:13 hiro-kobayashi: Additionally, how does blazar count the max of CUSTOM_RESOURCE in host reservation? 09:34:53 hiro-kobayashi: Have you already investigated whether 1) is sufficient or a combination of 1) and 3) would be necessary to support REQ3? 09:35:17 masahito: good point! I thinks blazar should not (cannot) count max 09:36:14 bertys_: Yes. But a patch is needed. https://review.openstack.org/#/c/561842/ 09:37:24 This patch enables update of the usage of nested resources when launching an instance. 09:37:29 hiro-kobayashi: Okay, thanks for sharing this info 09:38:10 nice. 09:39:26 I added info to the etherpad page 09:42:25 okay, another comment for the design? 09:42:39 if nothing, let's move on to next. 09:42:40 hiro-kobayashi: thanks, I need to catch up again with nested RPs ); Things are moving fast now 09:45:44 #topic AOB 09:46:08 Does someone have topics to share/discuss? 09:47:22 I'll share my update. 09:47:48 I pushed the patches for multi-freepools and flavor-extra-specs. 09:48:02 multi-freepools: https://review.openstack.org/#/c/564138/ 09:48:19 flavor-extra-specs: https://review.openstack.org/#/q/status:open+project:openstack/blazar+branch:master+topic:bp/flavors-extra-specs 09:48:47 Please review the patces. 09:48:52 That's from my side. 09:49:01 Great! 09:49:52 My test env is down now, so I'll review and test them as soon as it goes up. 09:50:38 thanks 09:50:57 I have reached out to Jean-Philippe (Ansible PTL) yesterday and there are few patches related to Blazar under review now e.g. https://review.openstack.org/#/c/566541/. Looks good so far 09:52:07 I am also planning to attend OpenStack PTG in Denver. Tentatively from Sep. 11 to 14 09:52:28 Good to hear. 09:54:11 anything else? if nothing, let's finish the meeting early. 09:54:26 Nothing from my side 09:56:23 OK, thanks everyone! 09:56:33 thanks! bye 09:56:35 #endmeeting