08:00:30 <ramishra> #startmeeting heat
08:00:31 <openstack> Meeting started Wed Oct  5 08:00:30 2016 UTC and is due to finish in 60 minutes.  The chair is ramishra. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:00:32 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:00:34 <openstack> The meeting name has been set to 'heat'
08:00:44 <ramishra> #topic roll call
08:00:49 <pas-ha> o/
08:01:08 <skraynev> o/
08:01:22 <prazumovsky> hi!
08:01:37 <stevebaker> hai
08:01:57 <ricolin> 0/
08:02:24 <therve> Heya
08:03:09 <ramishra> #topic adding items to agenda
08:03:19 <ramishra> #link https://wiki.openstack.org/wiki/Meetings/HeatAgenda#Agenda_.282016-10-05_0800_UTC.29
08:03:44 <stevebaker> memory usage
08:05:10 <ramishra> stevebaker: added that one
08:05:19 <ramishra> so we have just 2 topics
08:05:40 <ramishra> #topic heat summit sessions
08:05:56 <ramishra> #link https://etherpad.openstack.org/p/ocata-heat-sessions
08:06:10 <ramishra> should we wait for more time for any more proposals
08:06:15 <ramishra> or finalize them
08:06:36 <ramishra> I've updated the etherpad with what we had in that.
08:06:57 <ramishra> stevebaker: any topics from you on the client stuff?
08:08:34 <ramishra> therve: I've added a specific session on large stacks to cater to the magnum team.
08:08:39 <therve> ramishra, +1
08:08:53 <stevebaker> ramishra: could we finalise next week? I'm on leave until tomorrow
08:08:54 <therve> We'll talk tripleo as well presumably
08:09:09 <ramishra> stevebaker: sure
08:09:57 <ramishra> I've added a topic for validation improvements again, we can discuss about translation in that, if we want
08:10:21 <therve> Possibly, yeah. It may be worth another session though
08:10:27 <therve> We're missing one, too :)
08:10:56 <ramishra> though prazumovsky would not be around to participate.
08:11:32 <therve> It still worth having the discussion
08:12:27 <ramishra> we should start working on the validation improvements this cycle;)
08:12:57 <ramishra> therve: yeah I'll add that, if there is nothing form stevebaker, else we can club it with validation stuff.
08:13:04 <therve> OK
08:13:57 <ramishra> ok should we move on now?
08:14:11 <ramishra> #topic memory usage
08:14:59 <stevebaker> I've been out of the loop for the past 10 days, but things are happening here https://bugs.launchpad.net/heat/+bug/1626675
08:15:01 <openstack> Launchpad bug 1626675 in heat "Further memory usage issues with big stacks" [Critical,In progress] - Assigned to Zane Bitter (zaneb)
08:15:15 <ramishra> many of us are working on it. mostly therve and zaneb. therve, probably therve you can provide an update?
08:15:29 <therve> Sure
08:15:55 <therve> I think we spent a good chunk of time looking for a specific issue
08:16:00 <therve> But there doesn't seem to be one
08:16:10 <stevebaker> I wonder if we could use __slots__ elsewhere, like Stack https://review.openstack.org/#/c/382068/
08:16:10 <therve> Mostly a bunch of things adding up
08:16:34 <therve> slots may improve a bit, but it's minor
08:16:47 <therve> https://review.openstack.org/381588 is the latest that shaved about 10% for me
08:16:54 <therve> https://review.openstack.org/380542 is a good one too
08:17:43 <therve> I'll check https://review.openstack.org/382038 after the meeting
08:17:49 <ramishra> zaneb has few patches too
08:18:12 <therve> shardy wanted to do a server-less tripleo deployment
08:18:27 <therve> That would help testing a lot, as it takes about 30mins for me to check a patch
08:18:38 <therve> If we don't have servers that could be much faster
08:18:43 <ramishra> that would be great.
08:18:51 <therve> We need something that can fake a server, and signal data back
08:19:54 <stevebaker> therve: a dedicated integration test against a fake nova driver methings
08:20:00 <stevebaker> methinks
08:20:03 <therve> stevebaker, That's not enough
08:20:12 <therve> fake nova driver won't signal back
08:20:46 <stevebaker> yes, the integration test signals back
08:21:38 <therve> Right, but I don't want integration test
08:21:52 <therve> I want to use "openstack overcloud deploy"
08:22:19 <stevebaker> ah, and a standalone thing which signals
08:23:57 <therve> Yeah I don't really know yet
08:24:22 <therve> Something like TestResource but with the ability to signal data. Specified in a file maybe?
08:24:28 <stevebaker> yeah, it could even be interactive
08:24:30 <therve> That could be a memory hit though
08:24:36 <stevebaker> i'll have a think
08:25:37 <ramishra> so, we don't yet know how the memory consumption suddenly increased(in newton) for Tripleo as reported in the bug, right?
08:25:55 <therve> ramishra, I think there is not *a* reason
08:26:05 <therve> There are plenty of small reasons :)
08:26:18 <therve> One being simply that tripleo does more things
08:26:18 <ramishra> therve: all in newton:)
08:26:36 <therve> ramishra, One interesting test would be to check tripleo from early newton
08:26:40 <therve> To confirm that
08:27:05 <stevebaker> its a bit tricky, heat and tripleo-heat-templates evolved in parallel
08:27:43 <therve> I mean, old tht ought to work on heat master though
08:27:45 <ramishra> yeah, zeneb was probably suggesting it might be the thts or some process consuming all memory after they increased it?
08:27:48 <therve> Unless we have a regression
08:28:28 <therve> yaql is definitely among the culprits
08:28:49 <therve> They had to increase its memory limit, but only to 100k AFAIU
08:30:39 <ramishra> yeah, I've seen that discussion too.
08:32:27 <ramishra> yeah, we'll continue looking for improvements.
08:32:51 <ramishra> should we move on or anyone has any more to discuss?
08:33:14 <therve> I'm good
08:33:37 <stevebaker> I have nothing to add
08:33:45 <ramishra> #topic open discussion
08:34:45 <ramishra> btw, we got the stable:follows-policy patch merged https://review.openstack.org/#/c/379114/
08:35:25 <ramishra> Now we have the tag for both heat and python-heatclient
08:35:41 <therve> Cool
08:36:04 <stevebaker> +1
08:36:19 <ricolin> so we will get +1 here?
08:36:19 <ricolin> https://www.openstack.org/software/project-navigator/
08:36:56 <ramishra> ricolin: I assume so, when the newton one is available/
08:37:15 <therve> It is already there it seems
08:37:26 <ricolin> cool :)
08:39:52 <ramishra> yep, anything else?
08:39:54 <stevebaker> we could probably have a crack at ops-sdk-support and asserts_support-rolling-upgrade in this cycle
08:40:35 <ramishra> stevebaker: zaneb has suggested a summit session to discuss what we can do get those.
08:40:39 <ramishra> I've added that one
08:40:44 <stevebaker> +1
08:40:49 <ramishra> I think it's worth to discuss that
08:42:20 <ramishra> ok, anything else?
08:42:44 <ramishra> thanks all for joining
08:42:50 <stevebaker> \o
08:42:51 <ramishra> #endmeeting heat