14:00:23 #startmeeting watcher 14:00:24 Meeting started Wed May 31 14:00:23 2017 UTC and is due to finish in 60 minutes. The chair is alexchadin. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:25 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:28 The meeting name has been set to 'watcher' 14:00:39 Hello! 14:00:46 Hi 14:00:46 morning 14:01:32 Today we have the following agenda: https://wiki.openstack.org/wiki/Watcher_Meeting_Agenda#05.2F31.2F2017 14:01:53 #topic Announcements 14:02:10 #info One week left till pike-2 release 14:03:06 please, spend your time on high priority BPs during your review process 14:03:20 o/ 14:03:34 #info OpenStack Foundation proposed to use Storyboard instead of Launchpad 14:03:42 #link https://storyboard.openstack.org/ 14:03:57 Foundation tests new service named storyboard 14:04:03 lakerzhou 14:04:49 alexchadin: when will we be switching? 14:04:51 it uses auth of launchpad, but stores BP and bugs in different things 14:05:36 sballe_: I want Watcher team to play around with this site and we will decide whether we move to storyboard or not by next Wednesday 14:05:49 #link https://storyboard-dev.openstack.org/ 14:05:50 alexchadin: that's a big change if it gets accepted 14:06:19 here is test version to crush :D 14:06:46 vincentfrancoise: yes, so we need time to look at it 14:06:59 and to make right choice 14:07:14 #link https://storyboard-blog.io/ 14:07:46 here is description of storyboard and why Foundation wants to migrate to it 14:08:53 please, read above links once you have some free time 14:09:19 as vincentfrancoise said, it is really big change and we need to compare all pros and contras 14:10:10 alexchadin: +1 14:10:45 well, any other announcements from you? 14:11:35 great, let's move on 14:11:43 #topic Review Action Items 14:11:53 Watcher specs 14:12:02 Add spec for Supporting HA for background jobs in Watcher needs final core from Susanne #link https://review.openstack.org/#/c/447996/ 14:12:15 done 14:12:27 sballe_: if you +2 it, could you please send it to workflow? 14:13:02 done 14:13:10 sballe_: thanks! 14:13:24 Noisy Neighbor Strategy needs reviews (Hidekazu's comment is still valid) #link https://review.openstack.org/#/c/398162/ 14:14:06 psdhemdi is moving but he will be back next week and will upload a new spec 14:14:26 * sballe_ meant pshedimbi 14:14:30 sballe_: sure 14:14:48 #action pshedimbi submit new PS on https://review.openstack.org/#/c/398162/ 14:15:03 Support workload fingerprinting needs reviews #link https://review.openstack.org/#/c/448396/ 14:16:10 I ran out of time regarding review but will try this week 14:17:01 #action sballe_ review https://review.openstack.org/#/c/448396/ 14:17:29 Add spec for services-versioned-notifications-api bp needs final core review #link https://review.openstack.org/#/c/445817/ 14:18:00 sballe_: it would be good if you review it too 14:18:16 ok will do 14:18:24 sballe_: it is already accepted by me, vincentfrancoise, hidekazu and licanwei 14:19:23 alexchadin: 14:19:24 done 14:19:24 sballe_: since it has the same structure as other notification-related specs, it won't take a lot of time to review it 14:20:51 Add Volume migrate action spec needs reviews #link https://review.openstack.org/#/c/468817/ 14:21:52 hidekazu has submitted it recently, it is part of cinder related commits 14:22:41 nice! 14:22:42 vincentfrancoise: could you please review it? 14:23:12 alexchadin: add me on it, I'll do my best :) 14:24:05 alexchadin: I wanted to do some more reviews before the meeting but a meeting I wasn't aware took the time slot away from me... 14:24:26 vincentfrancoise: no worries 14:24:41 #action vincentfrancoise review https://review.openstack.org/#/c/468817/ 14:25:08 Build baremetal data model in watcher needs new PS #link https://review.openstack.org/#/c/448016/ 14:25:40 Yumeng__: have you seen Bin Zhou's comments? 14:26:02 Yep, I have seen them 14:26:10 I will update tomorrow 14:26:26 Since I just come back from holiday 14:26:56 Yumeng__: great! 14:27:17 #action Yumeng__ submit new PS on https://review.openstack.org/#/c/448016/ 14:27:43 Add new actions "power on" and "power off" in Watcher needs new PS #link https://review.openstack.org/#/c/435436/ 14:28:02 Yumeng__: here is the same :) 14:28:08 Ok 14:30:29 #action Yumeng__ submit new PS on https://review.openstack.org/#/c/435436/ 14:31:15 Watcher 14:31:30 alexchadin: Yumeng__: I just added a comment on https://review.openstack.org/#/c/448016/ to explain how we use notifications to build CDMs so that lakerzhou can see they are not really mandatory although highly encouraged 14:32:51 vincentfrancoise: thank you 14:33:18 Cancel Action Plan needs reviews #link https://review.openstack.org/#/c/458492/ 14:34:08 I will try to review it 14:34:27 #action alexchadin review https://review.openstack.org/#/c/458492/ 14:34:43 Watcher Install guide needs reviews #link https://review.openstack.org/#/c/463281/ 14:35:33 sballe_: could you please review it 14:35:34 ? 14:35:40 vincentfrancoise: thanks! 14:36:22 alexchadin: I am running out of time. I need to make sure I answers Bin Zhou's comments in our patch on noisy neighbor 14:36:37 I really want it merged as soon as possible 14:37:17 sballe_: okay :) 14:37:39 alexchadin: quick question about the lack of specs core reviewers since acabot is not here anymore, shouldn't we have all/most watcher core-reviewers as specs core-reviewers too so we end up with 1 core from each represented company? 14:38:43 vincentfrancoise: I plan to have talks with everyone from watcher core-reviewers this week 14:39:00 alexchadin: oh ok then :) 14:39:11 vincentfrancoise: I was talking about it during last meeting :) 14:39:39 alexchadin: yeah I saw the subject popped up although I didn't read it all 14:39:52 Yumeng__: do you have time to review install guide? 14:40:52 alexchadin, I will review the install guide 14:41:05 lakerzhou: thank you! 14:41:12 Yes, I can make time to do. 14:41:22 #action Yumeng__ lakerzhou review https://review.openstack.org/#/c/463281/ 14:41:51 Noisy Neighbor Strategy needs reviews #link https://review.openstack.org/#/c/454359/ 14:42:05 alexchadin: just answered Bin Zhou 14:42:44 I think the issue here is that not all noisy neighbor algorithm will server everybody 14:42:58 This algorithm is helping Intel in several ways. 14:43:18 It solves our issue and help illustrate how ll3 metric can be used 14:43:31 I agree that the code needs to be divided up in subroutines, etc. 14:43:49 but I am not sure I agree on questions around the algorithm 14:43:52 My main concern is around the decision of the dest_server 14:44:28 lakerzhou: I saw that. we are chosing the server with the least amount of cache occupancy which is what the algorithm focuses on 14:44:33 sballe_: Maybe a solution would be to add a description in which case it works and the assumptions it makes following the templating we used for other strategies like workload_balance.py 14:44:33 other than that, it is just code format, and code comments 14:45:06 so we migrate the aggressor to the server with the least amount of last level cache occupancy 14:45:38 vincentfrancoise: let me look at thta 14:47:01 vincentfrancoise: I added this as a comment to the code and pshembi will address it 14:48:38 lakerzhou: sballe_ vincentfrancoise nice work :D 14:49:01 vincentfrancoise: I agree about description 14:49:18 +1 14:50:07 * sballe_ :-) 14:50:15 +1 14:50:31 lakerzhou: every strategy (which is initiated by company) solves the issues which company faces 14:51:07 alexchadin: +1 14:51:25 lakerzhou: it may be not acceptable for other companies, but we are trying to provide wide-range strategies 14:52:22 lakerzhou: alexchadin: this is also why you have the distinction between goals and strategies: the former is generic the latter is more tied to the context of a company 14:52:35 +1 14:53:06 right 14:53:15 alexchadin, it is not realistic for one solution to fix all problems. Where should a user find the context of the solution? 14:53:43 A user need to know the risk if he/she wants to take the solution 14:54:33 lakerzhou: Whihc is why the description suggested by vincentfrancoise is a great idea 14:54:34 lakerzhou: user should find the goal that should be solved and look for goal's strategies, their descriptions and algorithm details 14:55:09 I am totally onboard with the description suggestion 14:55:26 lakerzhou: we already have description section for strategies 14:55:32 let me find a link... 14:56:00 lakerzhou: https://github.com/openstack/watcher/tree/master/doc/source/strategies 14:56:03 thanks, alex 14:57:06 lakerzhou: you can find a link to strategy specification inside of its documentation, section Algorithm 14:57:41 Excuse me, but I need to go 14:57:58 alexchadin: time's almost over anyway ;) 14:57:59 we have 3 minutes left so I propose to end up this meeting 14:58:05 I don't see one for noisy neighbor, so will we have one there? 14:58:07 ok make sense 14:58:12 bye and have a great day 14:58:28 lakerzhou: you are welcome on openstack-watcher channel :) 14:58:35 lakerzhou: yes, the answer to your question will be to write one of those ;) 14:58:47 sure, thanks all 14:58:51 bye 14:58:55 thank you for your attention 14:59:09 good luck and have a nice day 14:59:10 bye 14:59:16 #endmeeting