17:00:35 <boris-42> #startmeeting Rally
17:00:36 <openstack> Meeting started Tue Jun 17 17:00:35 2014 UTC and is due to finish in 60 minutes.  The chair is boris-42. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:37 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:40 <openstack> The meeting name has been set to 'rally'
17:00:48 <boris-42> msdubov hughsaunders marcoemorais1 ping
17:01:14 <hughsaunders> boris-42: hey
17:01:33 <andreykurilin> hi all
17:01:39 <k4n0> Hi Rally o/
17:02:54 <boris-42> so let' just wait a sec
17:02:58 <marcoemorais1> hey
17:04:19 <boris-42> okay let's just start
17:04:30 <boris-42> #topic hackings
17:04:39 <boris-42> Olga made a quite large patch
17:04:58 <boris-42> https://review.openstack.org/#/c/100599/
17:05:04 <boris-42> I think we should merge it ASAP
17:05:21 <boris-42> Cause it is blocking next patch
17:05:28 <boris-42> https://review.openstack.org/#/c/96833/
17:05:47 <boris-42> that contains a lot of updates in different projects
17:06:02 <boris-42> Thoughts ?
17:06:19 <k4n0> first patch or second?
17:06:48 <boris-42> k4n0 first & second we need both
17:06:54 <boris-42> without first we are not able to merge second
17:07:12 <boris-42> k4n0 second patch syncs rally requirements with requirements of rest openstack
17:07:24 <k4n0> then we should quickly review first patch
17:08:22 <boris-42> k4n0 yep
17:08:28 <boris-42> hughsaunders marcoemorais1 ^
17:08:42 <boris-42> guys you are core you should have your opinion on this as well lol
17:09:30 <marcoemorais1> boris-42: yes it is a big patch, need to look through it
17:09:34 <hughsaunders> Does look like a sizeable patch, will have to have  a read
17:09:37 <hughsaunders> marcoemorais1: snap
17:10:32 <boris-42> hughsaunders msdubov  okay good
17:10:45 <boris-42> #topic Keystone V3 support
17:10:56 <boris-42> Ok we find the guy  that will help us with this stuff
17:11:00 <boris-42> it's quite tricky
17:11:19 <k4n0> o/
17:11:24 <k4n0> can i help?
17:11:26 <boris-42> https://review.openstack.org/#/c/100579/
17:11:35 <boris-42> ^ k4n0  you can help if you review this
17:11:42 <k4n0> ok
17:11:59 <boris-42> So the idea is to create wrapper
17:12:04 <boris-42> on top of keystone client
17:12:14 <boris-42> so we will unify work with keystone for rest of rally
17:12:34 <boris-42> and it will detect automatically what version we are using and use coressponding API
17:12:42 <boris-42> k4n0 hughsaunders marcoemorais ^ thoguths ?
17:13:40 <marcoemorais> boris-42: interesting seems potentially useful even outside of rally
17:13:53 <k4n0> Is this a problem of deciding which keystone api to use? (https://github.com/openstack/heat/blob/master/heat/common/heat_keystoneclient.py)
17:14:20 <boris-42> k4n0 nope
17:14:36 <boris-42> k4n0 it's problem that in keystone v3 we have projects and in keystone v2 we have tenants
17:14:45 <boris-42> k4n0 they are similar but with different names...
17:15:01 <boris-42> and different api
17:15:20 <k4n0> ok, and what do we need to use?
17:15:43 <hughsaunders> k4n0: probably need compatibility with both for now
17:16:05 <k4n0> I see, so we need to lazy load the keystone client based on context
17:16:42 <boris-42> k4n0 if we specified in rally to use v3
17:16:54 <boris-42> k4n0 rally should still work+)
17:17:02 <boris-42> k4n0 at this point it doesn't work
17:17:09 <k4n0> I think heat does something similar (https://github.com/openstack/heat/blob/master/heat/common/heat_keystoneclient.py#L541)
17:17:22 <k4n0> You can specify default keystone backend.
17:17:31 <boris-42> k4n0 it's mess
17:17:37 <boris-42> k4n0 and not related to our task
17:17:39 <k4n0> ok
17:17:44 <boris-42> k4n0 I don't need delete_stack_domain_project
17:17:47 <boris-42> -)
17:17:52 <k4n0> I know :)
17:18:08 <boris-42> k4n0 so we will just unify basic operations that we are using
17:18:14 <boris-42> k4n0 that's all
17:18:18 <k4n0> ok
17:18:32 <boris-42> #topic Tempest Improvements
17:18:41 <boris-42> k4n0 could you share with your work and explain to others?
17:19:04 <k4n0> Yes
17:19:19 <k4n0> https://review.openstack.org/#/c/100481
17:19:28 <k4n0> I am removing dependency on subunit2junitxml
17:19:45 <k4n0> And parsing the raw subunit stream given by "testr" to json
17:19:50 <k4n0> basically subunit2json
17:20:36 <k4n0> questions?
17:21:18 <k4n0> https://blueprints.launchpad.net/rally/+spec/tempest-subunit-to-json
17:21:30 <boris-42> k4n0 agree actually with Andreay
17:21:37 <boris-42> k4n0 about location of subunit2json module=)
17:22:00 <k4n0> you want to invoke subunit2json from tool/subunit2json ?
17:22:09 <boris-42> k4n0 nope
17:22:12 <boris-42> I left comment
17:22:21 <boris-42> k4n0 put it deep inside tempest directory
17:22:33 <boris-42> k4n0 only tempest is using it and will use it
17:23:12 <k4n0> ok, so send subunit2json as patch to tempest? or copy it to tempest during installation from rally
17:23:14 <k4n0> ?
17:23:20 <andreykurilin> no
17:23:26 <boris-42> k4n0 send module
17:23:35 <boris-42> k4n0 move file in patch*
17:23:55 <k4n0> I dont understand
17:24:06 <boris-42> k4n0 file is in rally/subunit2json.py
17:24:16 <k4n0> ok got it
17:24:22 <boris-42> make it in rally/verification/veirifiers/tempst/subunit2json.py
17:24:28 <boris-42> only tempest is (and will use it)
17:24:29 <andreykurilin> +1
17:24:35 <boris-42> so it shouldn't be on top of everything
17:24:44 <boris-42> if we put everything in root
17:24:47 <boris-42> it will be mess
17:24:48 <boris-42> =)
17:25:03 <k4n0> ack
17:25:29 <k4n0> Ill make this and other change and submit new patch
17:25:38 <boris-42> k4n0 good
17:25:53 <andreykurilin> this module can be moved to subunit repo as one of subunit-filter :)
17:26:21 <boris-42> andreykurilin btw yes
17:26:47 <boris-42> andreykurilin but unit it is merged in it we will keep it in rally
17:27:05 <boris-42> lifeless ^ we made subunit2json.py
17:27:07 <andreykurilin> of course
17:27:18 <boris-42> lifeless are you interesting in getting it in subunit?
17:29:12 <boris-42> #topic ready to merge patch =)
17:29:38 <andreykurilin> boris-42, imo,  we need to send a letter to openstack-dev and discuss this filter
17:29:50 <boris-42> https://review.openstack.org/#/c/92873/12
17:30:07 <boris-42> andreykurilin k4n0 somebody of you guys can make this
17:30:15 <k4n0> andreykurilin: Ill send a mail to mailing list openstack-dev
17:30:24 <boris-42> hughsaunders msdubov marcoemorais guys could you take a look at patch above ^
17:30:32 * hughsaunders looks
17:30:34 <k4n0> #action send mail regarding subunit2json to openstack-dev
17:31:01 <andreykurilin> k4n0, ok
17:31:29 <boris-42> #action send mail regarding subunit2json to openstack-dev
17:32:10 <boris-42> #topic OpenStack cross service project profiler
17:32:19 <boris-42> Okay guys I finally finished work on osprofiler
17:32:27 <k4n0> great :)
17:32:28 <boris-42> https://github.com/stackforge/osprofiler
17:32:40 <boris-42> ^ There are just read me and small py33 fix left
17:32:47 <boris-42> and I'll cut release
17:33:04 <boris-42> After that I'll start integrating it in upstream
17:33:34 <boris-42> to integrate it in upstream we will need to put a couple of small patches
17:33:45 <boris-42> https://github.com/boris-42/nova/commit/9ebe86bf5b4cc7150251396cfb302dd05e89085d this is sample of nova integration
17:34:00 <boris-42> and this is required in ceilometer https://review.openstack.org/#/c/100239/
17:34:37 <boris-42> so as all patches are ready there is a bit bureaucracy
17:34:48 <boris-42> about blue prints and specs and so on
17:34:58 <k4n0> waiting for ceilometer patch to merge?
17:35:06 <boris-42> I need to make ceilometer-spec
17:35:15 <boris-42> to get approved this stuff
17:36:15 <k4n0> ok, so we need spec + patch for other projects also
17:36:25 <boris-42> k4n0 yep
17:36:29 <boris-42> k4n0 I will make one template
17:36:34 <boris-42> k4n0 and copy paste it lol
17:36:35 <boris-42> =)
17:36:37 <k4n0> ok
17:36:40 <boris-42> k4n0 but it will take some time
17:36:40 <k4n0> lol
17:36:50 <boris-42> and then there will be patch in rally
17:37:00 <boris-42> that integrates all this stuff so it will work out of box
17:37:00 <boris-42> =)
17:37:07 <k4n0> looking forward
17:37:11 <boris-42> and we will get traces for every iteration
17:37:23 <boris-42> in such format http://pavlovic.me/rally/profiler/
17:38:11 <boris-42> #topic open discussion
17:38:26 <boris-42> somebody would like to discuss anything ?
17:38:42 <hughsaunders> boris-42: good work on profiler
17:38:49 <k4n0> Rally release schedule (rpm, deb, pip)
17:39:20 <hughsaunders> k4n0: need versions before releases
17:39:25 <k4n0> https://blueprints.launchpad.net/rally/+spec/rally-distribution-packages
17:39:26 <k4n0> Yes
17:39:34 <k4n0> We need to discuss how we will version
17:40:21 <boris-42> yep this is good topic for discussion k4n0
17:40:30 <boris-42> when we are going to start versioning crap?
17:40:51 <boris-42> marcoemorais hughsaunders msdubov ^
17:41:48 <boris-42> I think there should be next topics
17:42:01 <hughsaunders> boris-42: as you've said previously its a trade off against velocity. We should start versioning when we intend to support the current version for a while, so UI, API, DB etc
17:42:02 <boris-42> are we going to support upgrades?
17:42:30 <boris-42> hughsaunders but we can just avoid supporting of upgrades
17:42:34 <boris-42> till moment X
17:42:35 <boris-42> =)
17:42:49 <boris-42> e.g. 0.*.* is without supporting upgrades
17:43:23 <boris-42> 1.*.* will be with support upgrades
17:43:37 <hughsaunders> I guess so.. we could go with the openstack cadence, release every 6 months, then that gives a time frame to work in
17:43:46 <boris-42> hughsaunders nope
17:43:58 <boris-42> hughsaunders I think it will be better to make short releases 1.5-2weeks
17:44:08 <boris-42> without supporting upgrades
17:44:28 <k4n0> how about going with tempest release cycle? the branchless one
17:44:35 <boris-42> until we will have stable API/UI/DB
17:44:47 <boris-42> k4n0 there is no more branches
17:44:57 <boris-42> k4n0 plus why we need to be aligned with them at all?
17:45:08 <boris-42> k4n0 imho versions should be done depending on changes in Rally
17:45:18 <k4n0> yes, bad idea, :)
17:45:22 <k4n0> just a thought
17:45:32 <boris-42> if we have a lot of changes for 1 week we can make version
17:45:53 <boris-42> if there is no changes we can wait 2 or even more weeks
17:46:08 <k4n0> how do we define "lot of changes" ?
17:46:14 <boris-42> i'll define=)
17:46:20 <k4n0> ok :)
17:46:21 <k4n0> works for me
17:46:27 <boris-42> depending on temperature on Mars
17:46:28 <boris-42> =)
17:46:34 <boris-42> so
17:46:38 <boris-42> about versioning
17:46:40 <k4n0> I am trying to get a rpm built internally in my office
17:47:01 <boris-42> x.x.* should be backward comp. fully
17:47:17 <boris-42> e.g. the same DB can be used with x.x.N and x.xN+K
17:47:31 <hughsaunders> boris-42: so push a tag to the repo every few weeks?
17:47:39 <boris-42> hughsaunders yep
17:47:59 <boris-42> x.N.* and x.N+1.* are not backward compability
17:48:06 <boris-42> and there is no upgrade for such change
17:48:24 <boris-42> cause I don't want to lose time at this point for upgrades
17:48:46 <boris-42> hughsaunders so actually we will make releases like python clients
17:48:56 <boris-42> hughsaunders they are doing it on demand
17:48:58 <boris-42> k4n0 ^
17:49:07 <hughsaunders> ok
17:49:12 <k4n0> +1
17:49:23 <boris-42> when we will be enough stable
17:49:30 <boris-42> when we move to 1.*.*
17:49:36 <boris-42> we will start supporting upgrades
17:50:25 <boris-42> k4n0 hughsaunders  so current versioning will be some kind of baby step to the full versioning
17:50:40 <k4n0> got it
17:51:13 <hughsaunders> boris-42: seems like a good idea, so we get the process going for releasing, but without the stability requirements as its 0.0.x
17:51:16 <boris-42> k4n0 could you put this in blueprint?
17:51:23 <boris-42> hughsaunders yep
17:51:31 <k4n0> boris-42: yes, on it
17:51:55 <boris-42> hughsaunders and when we get stability we will have all process about versioning and other stuff ready
17:52:02 <boris-42> hughsaunders and with good experience
17:52:02 <boris-42> =)
17:52:31 <hughsaunders> Sounds good
17:52:34 <boris-42> Other topics to dicuss?
17:52:59 <k4n0> nothing from my side
17:53:38 <boris-42> okay let's just finish meeting=)
17:53:44 <boris-42> Glad to see you guys
17:54:14 <k4n0> ok, cya. thanks all
17:54:17 <boris-42> #endmeeting