09:00:32 #startmeeting watcher 09:00:33 Meeting started Wed Nov 23 09:00:32 2016 UTC and is due to finish in 60 minutes. The chair is acabot. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:00:34 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:00:37 The meeting name has been set to 'watcher' 09:00:37 o/ 09:00:40 hi 09:00:44 o/ 09:00:47 hi guys 09:00:58 hi 09:01:01 europe is awake ;-) 09:01:32 agenda for today #link https://wiki.openstack.org/wiki/Watcher_Meeting_Agenda#11.2F23.2F2016 09:01:36 i guess the us is preparing for thanksgiving ;-) 09:01:47 seanmurphy : for sure 09:02:24 so lets start 09:02:31 #topic Announcements 09:02:47 #info python-watcher v0.32 & python-watcherclient v0.26 & watcher-dashboard v0.6 have been released for ocata-1 on November 18th 09:03:09 thx everyone for meeting the deadline 09:03:22 #info recap for Barcelona has been sent 09:03:30 #link http://lists.openstack.org/pipermail/openstack-dev/2016-November/107623.html 09:03:46 do not hesitate to answer the thread if something is missing 09:03:57 #info a discussion has been started regarding the way we should validate Watcher strategies 09:04:07 #link http://lists.openstack.org/pipermail/openstack-dev/2016-November/107402.html 09:04:32 this is an important issue so feel free to send your thoughts about it 09:04:39 on the ML 09:04:55 #info acabot didn't has the time to prepare the trello board for the meeting 09:05:03 sorry for that, I will do it next week 09:05:06 hello everyone 09:05:11 hello alex 09:05:21 #info acabot presented Watcher & our big tent journey on November 22nd @openstackdayfrance2016 09:05:49 I had 40 attendees 09:06:15 great! 09:06:23 #info your OSIC cluster request for testing Watcher has been rejected 09:06:45 acabot: why ? 09:06:48 they ask us to request a bare metal env instead of an openstack already deployed 09:06:56 well done on the openstack france presentation 09:07:19 and I dont really want to have to configure OpenStack for our tests 09:07:50 can we get some usage info from osic so we could estimate what gains could be made? 09:08:12 seanmurphy : I need to check that with sballe 09:08:28 seanmurphy : as I dont really know what metrics we can get 09:08:58 my main concern is to take 2 weeks to have an openstack ready with monasca deployed... 09:09:08 one idea that i thought might make sense - (I mentioned it in barca) - was if watcher could publish some usage data from some real systems as a driver for showing what gains are possible 09:09:11 and we can have a cluster for a month max 09:09:48 seanmurphy : yes this is something we planned to do before Barcelona, with the Intel cluster 09:10:12 seanmurphy : unfortunately it took 7 weeks to have openstack & watcher ready ! 09:10:31 haha - things always take longer than planned ;-) 09:10:47 seanmurphy : so we stopped the experimentation falling back on OSIC node with an openstack already deployed 09:11:20 seanmurphy : but as I said, they want us to deploy openstack as well... 09:11:35 acabot: how much servers? 09:12:03 alexchadin : we can have 132 nodes or 66 nodes cluster 09:12:47 I would suggest to work on improving watcher code until christmas and then ask for a testing env in january 09:13:23 I will check with Intel if they can handle the openstack deployment on the OSIC cluster 09:13:24 acabot: from my point of view, we have no choice, we have to run watcher within a medium cluster size, to be able to demonstrate the real gains with watcher 09:14:26 dtardivel : I agree but on bcom side, we wont be able to work on an openstack deployment on OSIC before Xmas 09:14:55 acabot: could we find some guys in our teams ready to help us to deploy a OS for our aims ? 09:15:00 #action acabot check with Intel if they could set up a new test environment for Watcher 09:15:27 dtardivel : I will try to find some 09:15:29 dtardivel: I have 2 years experience with deploying openstack 09:15:36 but only on very small cluster 09:15:42 4-5 machines 09:15:58 alexchadin: 8 machines for me :) 09:16:02 without automation tools like ansible, vagrant so on 09:16:12 I think configuring Newton on a 66 nodes env is not that easy ;-) 09:16:27 we need some tools to automate it 09:16:30 like fuel 09:16:34 s/Newton/Neutron 09:16:45 but it is in nearly dead state 09:17:13 let me ask for help next week 09:17:23 #topic Review Action Items 09:17:32 Watcher specs 09:17:39 automatic triggering of action plans for AUDIT should now be reviewed and merged 09:17:45 #link https://review.openstack.org/#/c/342473/ 09:18:12 alexchadin diga : can we merge it ? 09:18:18 I think yes 09:18:34 +2 from me 09:18:34 diga ? 09:18:50 hi 09:19:06 diga : can we merge your spec ? 09:19:14 nothing to add ? 09:19:14 acabot: yesm we can merge the spec 09:19:18 ok thx 09:19:36 #action acabot sballe review and merge https://review.openstack.org/#/c/342473/ 09:19:44 Define grammar for workload characterization needs a new PS 09:19:51 #link https://review.openstack.org/#/c/377100/ 09:20:04 thx seanmurphy for your comments 09:20:14 #action hvprash answer comments on https://review.openstack.org/#/c/377100/ 09:20:24 Define when an action plan is stale/invalid needs final reviews from cores 09:20:30 #link https://review.openstack.org/#/c/386293/ 09:20:54 I think this spec looks good 09:21:03 anyone wants to review it before merging ? 09:21:25 #action acabot sballe review and merge https://review.openstack.org/#/c/386293/ 09:21:32 +1 09:21:47 #action hanrong review and merge https://review.openstack.org/#/c/386293/ 09:21:51 hi hanrong ;-) 09:21:59 hi, :-D 09:22:02 Audit tag in VM Metadata needs reviews 09:22:12 #link https://review.openstack.org/#/c/391374/ 09:22:37 vincentfrancoise did a great addition to this spec 09:22:45 anyone wants to review it ? 09:23:07 I think its now complete 09:23:10 acabot: i think vincentfrancoise forgot to update this spec with our latest comments. I will do it. 09:23:23 dtardivel : PS 5 ? 09:23:29 I'd like to review it 09:23:46 yes i think so. I will check. 09:23:58 #action alexchadin acabot review https://review.openstack.org/#/c/391374/ 09:24:06 Add improvements to the planner and workflow mechanisms needs reviews 09:24:13 #link https://review.openstack.org/#/c/385871/ 09:24:40 we talked about splitting this spec in 2 09:24:55 dtardivel : did you talked with alexchadin about it ? 09:25:16 acabot: I talked with vincent about it 09:25:18 i can review 385871 09:25:37 seanmurphy: I appreciate it! 09:25:37 acabot: no, this is the position of vincentfrancoise. 09:26:01 acabot: In this spec I've described only the first part of BP 09:26:32 alexchadin : ok so should we update the BP description and add a new one ? 09:26:33 alexchadin: so only the data model impact ? 09:26:44 acabot: the second one, about modifying Applier, would be in the next spec. 09:26:56 seanmurphy : let me check is the scope is ok before reviewing 09:27:03 dtardivel: yes, but we cannot merge it without the second part:) 09:27:51 alexchadin : lets discuss this on our channel 09:27:53 so, these two specs cannot exists without each other 09:28:34 #action dtardivel alexchadin acabot sync on openstack-watcher about https://review.openstack.org/#/c/385871/ 09:28:43 Moving to Watcher code 09:28:53 alexchadin: from vincentfrancoise point of view, we should be able to update the data model and the way the planner will handle new model to make the action plan, without impacting the applier 09:29:09 watcher-notifications-ovo & audit-versioned-notifications-api merged 09:29:23 Documentation for Outlet Temperature Based Strategy looks good and will be merged soon 09:29:31 #link https://review.openstack.org/#/c/397411/ 09:29:42 is it the last doc strategy ? 09:30:33 dtardivel : do we still have strategies without doc ? 09:30:42 acabot: not sure 09:30:51 acabot: let me check 09:31:04 #action dtardivel check if we have all our strategies documented 09:31:11 centralise config options need reviews 09:31:22 #link https://review.openstack.org/#/q/status:open+project:openstack/watcher+branch:master+topic:bp/centralise-config-opts 09:31:30 who wants to review this ? 09:31:40 +1 09:31:48 +1 09:31:59 #action dtardivel hanrong review https://review.openstack.org/#/q/status:open+project:openstack/watcher+branch:master+topic:bp/centralise-config-opts 09:32:39 check the state of action plan needs a new PS 09:32:45 #link https://review.openstack.org/#/c/391383/ 09:32:56 #action licanwei answer comments on https://review.openstack.org/#/c/391383/ 09:33:05 Watcher CLI 09:33:10 Add functional tests to watcherclient needs reviews 09:33:17 #link https://review.openstack.org/#/c/389181/ 09:33:20 +1 09:33:28 #action dtardivel review https://review.openstack.org/#/c/389181/ 09:33:45 #topic Blueprint/Bug Review and Discussion 09:33:57 #info Ocata-2 is our current target due December 12th 09:34:04 #link https://launchpad.net/watcher/+milestone/ocata-2 09:34:36 acabot: uniform airflow strategy is not documented 09:35:07 nishi's strategy 09:35:13 I will ask sballe 09:35:31 #action acabot ask sballe for documenting airflow strategy 09:36:04 for ocata-2, we have 9 BPs targeted 09:36:25 please update the delivery field if you already started implementing 09:36:48 hanrong : did you start https://blueprints.launchpad.net/watcher/+spec/dynamic-action-description ? 09:36:55 yes 09:37:12 https://review.openstack.org/#/c/401111/ 09:37:39 hanrong : ops sorry I missed it 09:37:51 I submitted today 09:37:55 ok ;-) 09:38:14 36 min ago ;) 09:38:54 need more discussion 09:39:08 ok 09:39:18 #topic Open discussions 09:39:27 hanrong: I will read your proposition 09:39:38 #undo 09:39:39 Removing item from minutes: 09:39:57 #action dtardivel review https://review.openstack.org/#/c/401111/ 09:40:08 #topic Open discussions 09:40:24 I dont have any topic to discuss today 09:40:31 anyone wants to open one ? 09:40:31 I can also do review for https://review.openstack.org/#/c/401111/ 09:40:45 #action Yumeng review https://review.openstack.org/#/c/401111/ 09:41:50 we have been setting up monasca to understand resource consumption 09:41:57 as sballe suggested at barca meeting 09:42:02 it’s a pita! 09:42:08 seanmurphy : :-) 09:42:13 brunograz put up a blog post on it 09:42:19 please share it 09:42:24 we’re now thinking that prometheus is a better option 09:42:38 #link https://blog.zhaw.ch/icclab/installing-monasca-a-happy-ending-after-much-sadness-and-woe/ 09:42:47 seanmurphy: hi:) 09:42:53 I heard also yesterday that prometheus is probably the best option today for monitoring 09:43:08 I read your post yesterday and leaved some comments 09:43:20 interestingly, it seems to have little support so far for openstack 09:43:26 canonical are working on it afaik 09:43:43 anyway - it’s just a ‘by the way’ - not core to watcher, i guess but relevant 09:44:07 we did bounce around the idea of embedded a grafana dashboard into watcher on horizon 09:44:15 but we will see how doable it might be 09:44:27 ok 09:45:14 so thx everyone 09:45:24 have a good day/evening 09:45:25 promotheus is available on dockey hub: #link https://hub.docker.com/r/prom/prometheus 09:45:51 s/dockey/docker 09:45:54 dockey hub - i like it ;-) 09:46:24 yeah - it took 10 mins to set up prometheus to play with 09:46:28 vs days for monasca 09:46:37 # 09:46:49 * acabot not surprised ;-) 09:47:19 thx 09:47:22 #endmeeting