14:02:07 <andreykurilin__> #startmeeting Weekly Rally meeting
14:02:08 <openstack> Meeting started Mon Mar 21 14:02:07 2016 UTC and is due to finish in 60 minutes.  The chair is andreykurilin__. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:11 <openstack> The meeting name has been set to 'weekly_rally_meeting'
14:02:12 <amaretskiy> hi
14:02:21 <andreykurilin__> hi all
14:02:26 <rvasilets> o/
14:03:06 <redixin> hej
14:04:09 <andreykurilin__> we don't have agenda at https://wiki.openstack.org/wiki/Meetings/Rally again
14:04:34 <redixin> so nothing to discuss? =)
14:04:45 <andreykurilin__> no
14:04:49 <andreykurilin__> you are wrong)
14:04:59 <amaretskiy> release :)
14:05:09 <andreykurilin__> yeah
14:05:18 <redixin> & recursive atomic actions
14:05:35 <andreykurilin__> #topic New release
14:06:00 <andreykurilin__> so today we should have a new release
14:06:08 <andreykurilin__> based on out release schedule
14:06:45 <andreykurilin__> PS: out release schedule is located here - https://docs.google.com/spreadsheets/d/16DXpfbqvlzMFaqaXAcJsBzzpowb_XpymaK2aFY2gA2g/edit#gid=1993147046
14:07:13 <andreykurilin__> Do we have any blockers for release?
14:07:30 <rvasilets> #link https://docs.google.com/spreadsheets/d/16DXpfbqvlzMFaqaXAcJsBzzpowb_XpymaK2aFY2gA2g/edit#gid=1993147046
14:07:49 <rvasilets> I don't know any blockers)
14:08:12 <andreykurilin__> amaretskiy, redixin: ?
14:08:15 <rvasilets> If we stable enough then we can cut release)
14:08:30 <amaretskiy> i do not see blockers
14:08:38 <andreykurilin__> rvasilets: we are:)
14:08:55 <andreykurilin__> nice
14:08:59 <rvasilets> )
14:09:40 <andreykurilin__> so, I'll try to post release notes as soon as possible
14:10:04 <andreykurilin__> let's move to next topic
14:10:11 <andreykurilin__> #topic recursive atomic actions
14:10:20 <andreykurilin__> redixin: ^
14:10:28 <amaretskiy> #link https://review.openstack.org/#/c/279546/2/doc/specs/in-progress/improve_atomic_actions_format.rst
14:10:29 <ikhudoshyn> hi everybody
14:10:43 <andreykurilin__> ikhudoshyn: hi
14:10:55 <amaretskiy> i can add recursive atomic actions nesting to the schema in spec
14:11:14 <amaretskiy> but do we really need atomic actions nesting more then 1 level ?
14:11:21 <amaretskiy> *than
14:11:42 <andreykurilin__> #link https://allisonmaruska.files.wordpress.com/2015/10/we-need-to-go-deeper.jpg
14:12:44 <andreykurilin__> amaretkiy: I can not imagine cases when we need nested level more then 1
14:12:50 <boris-42> hi hihi
14:12:57 <andreykurilin__> hi boris-42
14:13:06 <boris-42> andreykurilin__: we need in complicated cases
14:13:45 <boris-42> andreykurilin__: like we would like to split first request and polling request
14:13:54 <boris-42> andreykurilin__: however it's the same action booting vm
14:14:07 <amaretskiy> okay, so add unlimited nesting?
14:14:37 <boris-42> amaretskiy: so to be honest for now I see only useful to have 1 level of it
14:14:45 <andreykurilin__> :)
14:15:04 <boris-42> amaretskiy: I would like to see the code that allows us to use 1 level (and that is simple to change and allow to use any levels)
14:15:08 <amaretskiy> i think that 1 level is OK because we can add deeper nesting easily
14:15:32 <boris-42> ok
14:16:59 <andreykurilin__> so we have an agreement on this topic
14:17:16 <amaretskiy> great, keep 1 level for now
14:17:22 <andreykurilin__> Any other topics to discuss?
14:17:42 <rvasilets> yes
14:17:49 <amaretskiy> I have one
14:17:57 <rvasilets> me too
14:17:58 <boris-42> andreykurilin__: yep
14:18:05 <andreykurilin__> why you didn't add them to agenda?)
14:18:12 <boris-42> andreykurilin__: didn't have time lol
14:18:21 <boris-42> andreykurilin__: and just got requirement
14:18:24 <andreykurilin__> :)
14:18:39 <andreykurilin__> ok, ravsilets was the first
14:18:45 <andreykurilin__> *rvasilets
14:18:51 <amaretskiy> actually I'm going to discuss a topic which kiran-r promised me to add to agenda, but he is absent, as well as a topic :)
14:18:51 <rvasilets> typing
14:18:55 <andreykurilin__> rvasilets: what is your topic?
14:19:22 <rvasilets> topic: rename atomic actions to something different
14:19:47 <rvasilets> I'm confused about using of atomic term in our case
14:20:00 <rvasilets> becaouse atom its the smallest part
14:20:09 <rvasilets> but we have nested levels
14:20:12 <andreykurilin__> #topic: rename atomic actions to something different
14:20:21 <amaretskiy> timing actions?
14:20:24 <boris-42> rvasilets: so your name actions
14:20:25 <rvasilets> so I suggest not to use atomic term
14:20:31 <rvasilets> just an action
14:20:32 <boris-42> rvasilets: ?
14:20:33 <rvasilets> for example
14:20:47 <ikhudoshyn> timed_action ?
14:20:49 <boris-42> rvasilets: yep that sounds actually reasonable and won't be hard to people to find it
14:20:55 <boris-42> ikhudoshyn: to long and not fancy
14:20:56 <boris-42> =)
14:21:01 <ikhudoshyn> )
14:21:01 <boris-42> too
14:21:02 <rvasilets> I suggest to rename atomic action across the project to just an action or, for example, activity action, measured action, measure.
14:21:35 <amaretskiy> just "timings" :)
14:21:36 <boris-42> rvasilets: andreykurilin__ amaretskiy so I believe actions is the best one, because it looks similiar to atomic_action
14:21:44 <rvasilets> yep
14:21:49 <ikhudoshyn> +1 for action
14:21:50 <rvasilets> agree
14:21:51 <andreykurilin__> I like timed action more
14:21:53 <amaretskiy> +
14:21:56 <boris-42> which won't produce any problems for old users
14:21:57 <andreykurilin__> :)
14:22:00 <amaretskiy> + for "action"
14:22:09 <boris-42> ok so seems we have agreement here
14:22:15 <rvasilets> yes
14:22:48 <rvasilets> Agreed: Rename atomic_actions to actions
14:22:59 <andreykurilin__> #agreement rename atomic_actions to actions
14:23:25 <andreykurilin__> amaretskiy: can you mention this rename at your spec?
14:23:47 <amaretskiy> okay, let's propose renaming in the spec
14:23:54 <andreykurilin__> nice
14:24:21 <andreykurilin__> just want to have a note somewhere about this rename
14:24:35 <amaretskiy> will do :)
14:24:38 <boris-42> andreykurilin__: is it BP enough?
14:25:14 <andreykurilin__> boris-42: it can be enough
14:25:55 <andreykurilin__> but I suppose, we can merge a spec by amaretskiy in near future so we aggregate info about refactoring atomics there
14:26:38 <amaretskiy> there will be at least 2 working items, so we need a BP
14:26:48 <andreykurilin__> ok
14:26:51 <andreykurilin__> :)
14:26:57 <andreykurilin__> let's move to the next topic
14:26:59 <amaretskiy> i will post a BP
14:27:13 <amaretskiy> my topic is about kiran-r question
14:27:26 <andreykurilin__> please share it:)
14:28:18 <amaretskiy> he asked me about use case - running 200 iterations over 20 tenants and expecting that each tenant will be used 10 times
14:28:28 <amaretskiy> however we do not have such balancing
14:28:33 <amaretskiy> we use random choice
14:28:38 <andreykurilin__> one moment
14:28:40 <amaretskiy> so thi sexpectation will not work
14:28:58 <boris-42> amaretskiy: this is not hard to do
14:29:01 <amaretskiy> the question is about balancing of tenants per iterations
14:29:02 <boris-42> andreykurilin__: implemenet
14:29:05 <amaretskiy> yes
14:29:11 <amaretskiy> this is relatively simple
14:29:14 <andreykurilin__> #topic possibility to balance usage of users
14:29:21 <boris-42> amaretskiy: we need just to add new parrameter to users/existing_users context
14:29:35 <andreykurilin__> we have an old implelmntation
14:29:39 <boris-42> andreykurilin__: strategy: random/round_robin
14:29:49 <andreykurilin__> #link https://review.openstack.org/#/c/229896/
14:30:47 <boris-42> andreykurilin__: so that has bugs
14:30:56 <boris-42> andreykurilin__: and bad UX it should be refactored before merged
14:31:15 <andreykurilin__> boris-42: you put -2 due to new type of runner
14:31:28 <andreykurilin__> which is not implemented yet:)
14:31:44 <amaretskiy> who will be responsible for this change?
14:32:01 <boris-42> andreykurilin__: I put -2 because you guys were trying to merge very bad patch
14:32:08 <boris-42> andreykurilin__: that blocks work on distributed runner
14:32:13 <andreykurilin__> boris-42: :)
14:32:33 <boris-42> andreykurilin__: so you tried to shoot your leg and I stopped that
14:32:36 <andreykurilin__> amaretskiy: If we leave suggestions, vponomaryov can finish it
14:32:59 <boris-42> andreykurilin__: There are already suggestions ...
14:33:02 <amaretskiy> okay, I will add a comment
14:33:12 <boris-42> amaretskiy: there is already  comment form me
14:33:16 <andreykurilin__> boris-42: I spent too much time for rally verify(Tempest), so it is normal for me to shoot my legs:)
14:33:29 <boris-42> andreykurilin__: amaretskiy guys you need to read all comments for all patch sets
14:33:46 <boris-42> andreykurilin__: amaretskiy and not skip them esepcially when they put -1
14:33:52 <amaretskiy> ok
14:34:14 <andreykurilin__> let's add suggestions how comments by boris-42 can be addressed:)
14:34:26 <boris-42> andreykurilin__: ok
14:35:09 <andreykurilin__> let's move to next topic
14:35:15 <andreykurilin__> boris-42: your turn
14:37:06 <andreykurilin__> boris-42: ping
14:37:07 <andreykurilin__> :)
14:37:43 <rvasilets> maybe he is typing?)
14:37:44 <boris-42> andreykurilin__: so "shaker to rally"
14:37:59 <andreykurilin__> #topic shaker to rally
14:38:07 <boris-42> redixin: are you around?
14:38:15 <redixin> yes
14:38:27 <boris-42> redixin: so basically shaker runs heat template after that triggers net tools
14:38:32 <andreykurilin__> #link https://github.com/openstack/shaker
14:39:26 <boris-42> redixin: it will be nice to generalize work that you did for MCV team and implement shaker functionallity
14:39:48 <boris-42> redixin: can you spend some time during this week and do analyze of what should be done and calulate estimates?
14:40:06 <redixin> boris-42: yea I was thinking about heat workload today, and have an idea
14:40:46 <redixin> we can make heat_stack context and scenario to do something with this context, like do heat.update or ssh("iperf")
14:41:23 <boris-42> redixin: so like in context we are running heat create
14:41:28 <boris-42> redixin: in scenario heat update
14:41:45 <redixin> I mean heat_stack will deploy something, and scenario will run something (heat_update or anything else)
14:41:57 <boris-42> redixin: yep
14:42:08 <redixin> run something with or without updating heat
14:42:09 <boris-42> redixin: so we will be able to use heat context to create any env
14:42:16 <boris-42> redixin: I like the idea
14:42:23 <redixin> like heat.update({some_key: self.context]
14:42:35 <redixin> like heat.update({some_key: self.context[iteration])
14:42:36 <boris-42> redixin: so can you do analyze of shaker and what should be done to implemenet
14:42:41 <redixin> sure
14:42:47 <boris-42> self.context["heat"]["blablabla"]
14:42:50 <boris-42> actually=)
14:43:11 <boris-42> but I got the idea
14:43:19 <redixin> we need to know current iteration number
14:43:38 <redixin> use iteration number in heat.update
14:43:49 <redixin> like heat.update({num_vms: iteration_number})
14:44:00 <redixin> and limit concurrency to 1
14:44:17 <redixin> so we can test heat stack with different number of some worker nodes
14:44:21 <andreykurilin__> redixin: we have a variable self.context["iteration"]
14:45:33 <boris-42> andreykurilin__: yep we have
14:45:52 <boris-42> andreykurilin__: it's add by scenario runner as far as I know
14:46:17 <andreykurilin__> yes, so we can determice current iteration number easily
14:46:22 <andreykurilin__> *determine
14:46:31 <redixin> right
14:50:09 <boris-42> okay
14:50:11 <boris-42> maybe next topic?
14:50:13 <boris-42> andreykurilin__: ^
14:50:25 <andreykurilin__> It looks like we don't have more topics
14:50:28 <andreykurilin__> :)
14:50:31 <rvasilets> https://github.com/openstack/rally/blob/master/rally/task/runner.py#L57
14:50:31 <boris-42> andreykurilin__: we have
14:50:53 <boris-42> andreykurilin__: topic: google doc 2 BP and back
14:51:02 <andreykurilin__> #topic google doc 2 BP and back
14:51:18 <boris-42> so the idea is to move all data from google doc to BP
14:51:20 <andreykurilin__> boris-42: you have 9 minutes for it:)
14:51:22 <boris-42> as well update it
14:51:38 <boris-42> and after that use script to generate google spreadsheets from launchapd
14:51:52 <boris-42> ikhudoshyn: ^ what is the status of this work
14:52:17 <ikhudoshyn> I started digging across the doc, it's huge
14:52:58 <ikhudoshyn> I didnt create any bp so far
14:53:39 <ikhudoshyn> but I don't think we need any scrip or smth for that
14:53:46 <ikhudoshyn> script
14:54:00 <ikhudoshyn> so my primary intent is to make it all manually
14:54:27 <ikhudoshyn> but i'm in doubts
14:55:10 <ikhudoshyn> not much by far
14:55:16 <ikhudoshyn> but that's all for now
14:56:02 <boris-42> ikhudoshyn: first time from google doc -> to BP you will have to do maunally
14:56:04 <boris-42> for 2 reaosns
14:56:15 <boris-42> 1) data is sometimes out of dated in both places
14:56:31 <boris-42> 2) it is super hard to do with scripts (working with LP is painful)
14:56:39 <boris-42> ok
14:57:01 <ikhudoshyn> yep, agree on both
14:57:52 <ikhudoshyn> but there are 60+ items in the doc so it won't happen super fast
14:59:18 <boris-42> ok
14:59:27 <boris-42> seems like we are out of time
14:59:38 <andreykurilin__> yeah
14:59:43 <andreykurilin__> we need to finish
14:59:52 <andreykurilin__> #endmeeting