08:04:33 <rakhmerov> #startmeeting Mistral
08:04:34 <openstack> Meeting started Wed Jul 31 08:04:33 2019 UTC and is due to finish in 60 minutes.  The chair is rakhmerov. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:04:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:04:36 <rakhmerov> vgvoleg: go ahead
08:04:37 <openstack> The meeting name has been set to 'mistral'
08:05:35 <vgvoleg> I am not sure where I should write notifications docs
08:06:01 <vgvoleg> Main features?
08:06:42 <rakhmerov> aaah.. let me think
08:06:54 <rakhmerov> so
08:07:31 <rakhmerov> I'd say "yes" but I'd suggest to have a separate page about it for now and insert only a link into "Main features"
08:07:43 <rakhmerov> because the current structure is doc is ugly
08:07:46 <rakhmerov> I don't like it
08:07:47 <vgvoleg> Also, there is a huge gap between what exists in mistral and what is described in the documentation
08:07:59 <rakhmerov> and we have plans to restructure it completely soon
08:08:12 <vgvoleg> where can I read this plans?)
08:08:28 <rakhmerov> I'm telling you :) It's mostly in our heads now
08:08:49 <rakhmerov> we also have some tickets but it's hard to describe everything that needs to be done in a ticket
08:08:54 <rakhmerov> it's a creative process
08:09:10 <rakhmerov> vgvoleg: Oleg, please either fix those mismatches or file tickets
08:09:19 <rakhmerov> I'd ask others to do it as well
08:09:27 <rakhmerov> we need to be tracking it somehow
08:09:30 <vgvoleg> I'd like to help, otherwise I will continue to support our internal docs, which is sad :D
08:09:48 <rakhmerov> no-no
08:09:53 <rakhmerov> it's counter-productive
08:10:00 <rakhmerov> we work on a community project
08:10:05 <rakhmerov> let's make it better
08:10:12 <rakhmerov> we have all freedom
08:10:28 <vgvoleg> I don't like to support our internal docs for sure :D
08:11:08 <rakhmerov> :)
08:11:17 <vgvoleg> what else we can do instead of one-page features description?
08:11:19 <rakhmerov> feel free to fix gaps
08:11:28 <rakhmerov> for now w/o changing the structure too much
08:11:35 <vgvoleg> because I am not sure if this a good idea or not
08:12:49 <rakhmerov> I recommend to create a separate page just as a temporary approach
08:13:04 <rakhmerov> because you're right, it's not clear now where to best add it
08:13:13 <rakhmerov> the structure is bad
08:13:40 <rakhmerov> so I think we need to create it somewhere (not too important for now where exactly) and then we'll find a better place for it
08:13:43 <rakhmerov> makes sense?
08:15:03 <vgvoleg> I think the perfect way is to implement good table of contents and to have one page per feature
08:15:22 <rakhmerov> sure
08:15:25 <rakhmerov> yes
08:17:27 <rakhmerov> anything else? :)
08:18:03 <rakhmerov> I'm investigating one simple (as I think) bug for now and then go back to improving scheduler
08:19:02 <vgvoleg> I dont have anything else
08:20:05 <rakhmerov> ok
08:20:08 <rakhmerov> let's finish then
08:20:09 <vgvoleg> oh
08:20:12 <rakhmerov> thanks vgvoleg
08:20:12 <vgvoleg> wait
08:20:15 <rakhmerov> yep
08:21:25 <vgvoleg> Do we have any guides about resources limits/scale replicas to hit performance goals?
08:24:09 <vgvoleg> Just something about what new customers want to know before using mistral
08:24:46 <vgvoleg> something in terms of "tasks per second" or smth like that
08:24:52 <rakhmerov> second..
08:25:07 <rakhmerov> vgvoleg: no
08:25:26 <rakhmerov> I'll tell you more, those properties change from cycle to cycle
08:25:40 <rakhmerov> towards being better, of course
08:27:00 <vgvoleg> sad
08:27:20 <rakhmerov> priorities, resources, as usual
08:27:38 <rakhmerov> we've been focus to make it better so far
08:28:06 <rakhmerov> you can work on such a guide btw :) You're experience enough already ;)
08:28:11 <rakhmerov> experienced
08:28:43 <vgvoleg> yes
08:28:46 <vgvoleg> i'll try
08:28:53 <vgvoleg> just with std.noop
08:30:10 <vgvoleg> without action overhead
08:30:17 <rakhmerov> yep
08:30:24 <rakhmerov> I can help you, of course
08:31:13 <rakhmerov> keep in mind that benchmarking is trickier than it seems at first look
08:31:31 <rakhmerov> you may want to read how good benchmarks are done
08:33:22 <rakhmerov> ok, let's close the meeting for now
08:33:26 <rakhmerov> #endmeeting