09:05:20 #startmeeting self-healing 09:05:21 Meeting started Wed Jun 5 09:05:20 2019 UTC and is due to finish in 60 minutes. The chair is aspiers. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:05:22 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:05:25 The meeting name has been set to 'self_healing' 09:05:35 #topic report from billing initiative meeting 09:05:46 go for it :) 09:05:49 :) 09:06:19 Public Cloud SIG has started the "Billing Initiative" 09:06:30 I have attended the last two meetings 09:06:42 http://eavesdrop.openstack.org/meetings/publiccloud_wg/2019/publiccloud_wg.2019-05-23-14.00.log.html 09:06:50 http://eavesdrop.openstack.org/meetings/publiccloud_wg/2019/publiccloud_wg.2019-05-28-14.05.log.html 09:07:20 the discussions so far focused on collecting and storing the required measurements 09:08:02 I was promoting the idea of instrumenting the code of OpenStack services 09:08:43 OK 09:08:45 I think it's in a long term the best approach both for billing and monitoring in general 09:09:37 guys have not been very enthusiastic about it though 09:09:42 Does that tie in with what we discussed in Denver with Dirk / Ben / others about exposing internal service metrics via endpoints? 09:10:04 yes, that's the same approach 09:10:19 OK 09:10:37 I just had a crazy idea for the future :) 09:11:01 Billing could offer refunds if monitoring sees that users are impacted by outages 09:11:21 that kind of ties billing with self-healing 09:11:23 nice use case :) 09:13:09 anyway, they pointed out that the implementation will take long time and were not optimistic if it can be finished at all 09:13:43 Mohamed Nasser suggested writing Prometheus exporters instead 09:13:56 OK 09:15:00 I think, instrumenting the code is not much effort, it just requires coordination with other projects 09:15:08 Yeah 09:15:15 and will pay off in long term 09:15:28 Maybe write a spec? 09:15:31 because instrumentation will live with the code 09:15:57 and is able to collect more data than `black box` monitoring with exporter 09:16:02 Yeah 09:16:10 A spec would be a good way to help people understand your vision 09:16:21 You could give some example code to show how it would work 09:17:00 yes, I think I should prioritize it on my list 09:17:54 that's all from me I guess 09:18:02 Not sure where the best place for the spec is 09:18:57 if we want to provide oslo library, probably there, but I'm not sure if we need one 09:19:46 self-healing could be alternatively 09:20:07 yup 09:20:36 well you can draft the spec and submit it somewhere 09:20:44 if it's the wrong place it's easy to move :) 09:20:57 correct 09:23:02 alright, thanks for reporting about that 09:23:18 #topic AOB 09:23:32 I don't have any updates except I still need to send a report about Denver :-( 09:23:38 anything else from your side? 09:24:06 ideas for session proposals for Shanghai? 09:24:44 oh, good point 09:25:01 we should try the same one we submitted last time with Ifat 09:25:30 yes, that might work 09:27:17 I don't have anything else 09:27:20 nor me 09:27:25 when's the deadline for Shanghai? 09:27:27 July? 09:27:35 early July 09:27:37 OK 09:27:40 We have time :) 09:27:46 Maybe we can talk to the new Vitrage PTL about it 09:27:55 right 09:28:00 Cool 09:28:11 next two weeks I'm in vacation 09:28:17 I won't have time this week either 09:28:32 Do you still have the text of the submission? 09:28:45 yes, will find it 09:28:48 Maybe best to forward to him sooner, so he has time to think about it while you are away 09:28:52 Cool, thanks 09:29:22 Alright, thanks again and catch you soon! 09:29:31 thanks, bye 09:29:35 o/ 09:29:37 #endmeeting