14:02:19 <rvasilets> #startmeeting Rally
14:02:20 <openstack> Meeting started Mon May 16 14:02:19 2016 UTC and is due to finish in 60 minutes.  The chair is rvasilets. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:22 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:24 <openstack> The meeting name has been set to 'rally'
14:02:32 <andreykurilin> o/
14:02:33 <rvasilets> Hi to all
14:03:15 <rvasilets> boris-42, stpierre
14:04:30 <rvasilets> amaretskiy,
14:04:33 <boris-42> rvasilets: hi
14:04:36 <amaretskiy> hi
14:04:59 <rvasilets> Looks like our meeting will be short
14:05:18 <rvasilets> #topic Review request
14:05:39 <rvasilets> Okey the only topic is mine
14:05:45 <rvasilets> I want more review
14:06:15 <rvasilets> First https://review.openstack.org/#/c/283180/ its hanging from 27th april
14:06:25 <boris-42> rvasilets: everybody wants more review+)
14:06:49 <rvasilets> Second https://review.openstack.org/#/c/314536/ this is the fix to the high priority bug
14:07:28 <rvasilets> Third https://review.openstack.org/#/c/315657/ As I understand the only reasonable patch in future coming release
14:07:32 <rvasilets> eom
14:08:00 <andreykurilin> ok
14:08:07 <andreykurilin> I have several topics
14:08:15 <rvasilets> Okey
14:08:25 <andreykurilin> "usage of global-requirements"
14:08:52 <rvasilets> #topic usage of global-requirements
14:09:08 <boris-42> andreykurilin: so that's good topic
14:09:13 <andreykurilin> #link https://review.openstack.org/#/c/316794/
14:09:41 <boris-42> andreykurilin: however as far as I understand as far as we use devstack
14:09:48 <andreykurilin> boris-42: no
14:09:49 <andreykurilin> :)
14:09:49 <boris-42> andreykurilin: we have to use global requirments
14:10:06 <andreykurilin> I had a talk with jd__ today
14:10:18 <andreykurilin> gnocchiclient do not use g-r
14:10:34 <andreykurilin> and everything is ok
14:11:05 <andreykurilin> telemetry team has an 2 years experience without usage of g-r
14:11:19 <boris-42> andreykurilin: sooo
14:11:27 <boris-42> andreykurilin: than maybe it's time to think about it=)
14:11:32 <andreykurilin> yeah:)
14:12:15 <boris-42> andreykurilin: we would be able to do upper constraits as well
14:12:20 <andreykurilin> yeah
14:12:23 <andreykurilin> we need it
14:12:25 <andreykurilin> for releases
14:12:55 <andreykurilin> first of all, if we decided to abandon g-r for Rally, we will need to sync latest requirements(which are in g-r) with our list
14:13:14 <andreykurilin> we will need to implement several small jobs
14:13:41 <andreykurilin> but it should not be a hard task
14:14:18 <andreykurilin> I propose to think a bit about g-r. what are advantages and disadvantages of its usage
14:14:30 <andreykurilin> and discuss this topic at the next meeting
14:14:56 <boris-42> andreykurilin: so does ceilometer have this jobs?
14:15:01 <rvasilets> I though that if we are OPenStack that we should use g-r
14:15:17 <rvasilets> *we must
14:15:37 <andreykurilin> rvasilets: I think it was an old requirement which was rejected while moving to "big tent"
14:15:52 <andreykurilin> rvasilets: now, we have an ability to reject g-r
14:16:01 <andreykurilin> boris-42: I don't know
14:16:24 <rvasilets> Ou, really) Then I have nice improvements to Rally logic =)
14:16:37 <andreykurilin> boris-42: I think we can ask jd__ to share his workflow about requirements.
14:16:55 <rvasilets> Who is he?
14:16:59 <boris-42> rvasilets: andreykurilin btw I would prefer to remove pbr and testr
14:17:05 <andreykurilin> heh
14:17:13 <boris-42> rvasilets: andreykurilin and use directly setuptools and pytest
14:17:20 <boris-42> that will make rally work on windows
14:17:21 <rvasilets> That was one from my suggestions)
14:18:01 <andreykurilin> rvasilets: http://stackalytics.com/?user_id=jdanjou
14:18:31 <andreykurilin> boris-42: yeah, I think we will be able to reject pbr
14:18:46 <rvasilets> lol Core in: pbr
14:18:49 <andreykurilin> yeah:)
14:19:36 <boris-42> andreykurilin: so I think like not supporting windows is a good reason+)
14:19:40 <boris-42> andreykurilin: to make things simpler
14:19:49 <rvasilets> yes
14:20:20 <rvasilets> And we will be able to rempve some extra code
14:20:27 <rvasilets> like percentile
14:20:31 <rvasilets> or LockedDict
14:21:51 <andreykurilin> ok, let's think a bit more about g-r,  talk with redixin about implementation of new jobs
14:21:56 <boris-42> rvasilets: why do you want to remove Lockeddict?
14:22:15 <boris-42> rvasilets: to add more useless dependendinces ?)
14:22:24 <andreykurilin> boris-42: I think, rvasilets means usage of external libraries for such stuff
14:22:37 <boris-42> andreykurilin: that is very bad patterns
14:22:40 <boris-42> pattern
14:23:01 <boris-42> we should use as less as possible libs
14:23:07 <boris-42> and depend on less things
14:23:15 <rvasilets> boris-42, okey boss but for persentile its reasonable!
14:23:32 <rvasilets> there is better implementation though C in numpy
14:24:00 <boris-42> rvasilets: numpy btw is very very haevy package
14:24:09 <boris-42> rvasilets: and is it stream algorithm?
14:25:09 <rvasilets> Talking only about percentile: In numpy its faster
14:25:15 <rvasilets> No its not streaming
14:26:10 <boris-42> rvasilets: then it's bad non working for us algorithm
14:26:55 <rvasilets> boris-42, then numpy developer is not so clever
14:27:00 <andreykurilin> lol
14:27:15 <andreykurilin> let's move to the next topic
14:27:22 <boris-42> rvasilets: oh god
14:27:28 <boris-42> rvasilets: you need a car
14:27:35 <boris-42> rvasilets: but here is motocycle
14:27:41 <boris-42> rvasilets: but I don't need motocycle
14:27:53 <boris-42> rvasilets: oh developer of motocycle are not so clever
14:27:56 <boris-42> rvasilets: WTF man??
14:28:29 <boris-42> rvasilets: we can use only streaming algorithms otherwise, we won't scale
14:28:42 <boris-42> rvasilets: if they don't have such implementaiton, we can't use it
14:30:11 <rvasilets> Okey lets move
14:30:17 <rvasilets> what the topic?
14:30:30 <andreykurilin> glossary
14:30:34 <amaretskiy> +
14:30:44 <rvasilets> #topic Glossary
14:30:45 <andreykurilin> #link https://review.openstack.org/#/c/266389/
14:30:58 <andreykurilin> first of all I want ask all of you to review this patch
14:31:00 <amaretskiy> http://docs-draft.openstack.org/89/266389/10/check/gate-rally-docs/bc530ca//doc/build/html/glossary.htmlъ
14:31:04 <amaretskiy> http://docs-draft.openstack.org/89/266389/10/check/gate-rally-docs/bc530ca//doc/build/html/glossary.html
14:31:19 <andreykurilin> It describes most of base terms of Rally
14:31:53 <andreykurilin> While reviewing some terms, I found that we have at least one bad term
14:31:55 <boris-42> andreykurilin: i'll try to review it soon
14:32:07 <andreykurilin> "type" in task component
14:32:12 <rvasilets> #link http://docs-draft.openstack.org/89/266389/10/check/gate-rally-docs/bc530ca//doc/build/html/glossary.html
14:32:39 <andreykurilin> This entity relates only to arguments not to full task
14:32:59 <andreykurilin> Imo, we should rename it to something like arguments_type or param_type
14:33:15 <andreykurilin> Any ideas?
14:34:11 <boris-42> andreykurilin: nothing for now
14:34:34 <boris-42> andreykurilin: what I would prefer is to finish first refactoring of input task format
14:34:40 <rvasilets> I don't have ideas for now
14:34:42 <boris-42> andreykurilin: that is the source of evil
14:35:03 <boris-42> andreykurilin: btw I have a topic
14:35:03 <amaretskiy> porposal: rename rally.task.types.ResourceType -> rally.task.arguments.Argument
14:35:27 <boris-42> amaretskiy: I talked on firday to AppFormix
14:35:31 <boris-42> andreykurilin: ^
14:35:41 <andreykurilin> amaretskiy: sounds better
14:35:46 <boris-42> they are interested in 2 things
14:35:48 <andreykurilin> boris-42: they want to contribute to rally?
14:35:50 <andreykurilin> :)
14:35:52 <boris-42> aaS
14:36:11 <boris-42> & helping with making rally able to benchmark non openstack envs
14:36:15 <andreykurilin> cool
14:36:24 <amaretskiy> I'm being ready to start implementing AaS for last year :)
14:36:48 <andreykurilin> aaaS - anything as a service :D
14:37:35 <andreykurilin> boris-42: Do they need a spec for aas or some chat will be enough?
14:38:01 <rvasilets> We are all ready)
14:38:03 <amaretskiy> boris-42: lets just start implementing RaaS, I'm ready
14:38:40 <boris-42> amaretskiy: and what about services base?)
14:39:11 <amaretskiy> boris-42: I has reviewed this today
14:39:54 <boris-42> amaretskiy: okay maybe we should start
14:40:14 <amaretskiy> boris-42: I remember we have a spec
14:40:42 <amaretskiy> #link https://review.openstack.org/#/c/182245/
14:40:48 <amaretskiy> we have to revive it
14:42:17 <rvasilets> reborn of an old manuscript)
14:42:58 <boris-42> amaretskiy: yep
14:43:04 <boris-42> amaretskiy: so we need to complete it
14:43:20 <amaretskiy> boris-42: I take it
14:44:28 <rvasilets> Okey its more similar to the Open discussion  anything else related to Glossary?
14:44:46 <andreykurilin> nothing from me
14:45:07 <rvasilets> boris-42, amaretskiy ^
14:45:27 <boris-42> nothing from me
14:45:37 <amaretskiy> I think we have to merge base version of glossary to finally get it available in docs, and improve it later in another patches
14:45:56 <amaretskiy> nothing from me
14:45:59 <rvasilets> #topic Free discussion
14:46:37 <rvasilets> feel free to flood=)
14:49:39 <andreykurilin> it looks like we can finish meeting
14:51:24 <rvasilets> #endmeeting