17:01:08 <boris-42> #startmeeting Rally
17:01:08 <openstack> Meeting started Tue Mar 18 17:01:08 2014 UTC and is due to finish in 60 minutes.  The chair is boris-42. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:01:11 <openstack> The meeting name has been set to 'rally'
17:01:37 <boris-42> marcoemorais hughsaunders stannie meeting
17:03:28 <stannie> yes boris-42
17:03:47 <stannie> hi
17:04:17 <boris-42> stannie hey there=)
17:04:20 <boris-42> msdubov ping
17:04:27 <boris-42> penguinRaider ping
17:04:31 <penguinRaider> boris-42, pong
17:04:33 <boris-42> marcoemorais ping
17:05:38 <msdubov> boris-42 hi
17:07:57 <boris-42> so let start
17:08:01 <boris-42> our meeting stuff
17:08:55 <boris-42> #topic Benchmarking future steps
17:09:15 <boris-42> So stannie  you are improving perfromance of generic cleanup?
17:09:33 <hughsaunders> pong
17:09:52 <stannie> yes boris-42 I am going to work on CleanupContext
17:09:56 <stannie> haven't started yet
17:10:50 <stannie> the task is to add a special decorator for CleanupContext that will increase the performance by having a a generic cleanup
17:11:41 <stannie> e.g @cleanup(["service1", "service2"]) will only run cleanup for service1 and 2
17:11:45 <stannie> service1 can be nova
17:11:54 <boris-42> yep yep
17:11:54 <stannie> service2 e.g cinder etc
17:12:05 <boris-42> stannie btw Kun will work on another improvment
17:12:15 <stannie> ok so he is depending on my patch ?
17:12:19 <boris-42> nope
17:12:23 <stannie> ok
17:12:30 <boris-42> but he will improve as well generic clenaup
17:12:38 <boris-42> to cleanup only users that we actually used
17:13:04 <boris-42> e.g. if we have 10k users but run tests only 1k times (we don't need to cleanup all 10k users)
17:13:16 <msdubov> boris-42 Have you started the work on the StressRunner?
17:13:25 <boris-42> msdubov nope I didn't..
17:13:39 <msdubov> boris-42 Ok I'll take it then if you're not against
17:13:47 <boris-42> msdubov okay
17:14:01 <boris-42> stannie hmm so I think that everybody agrees with that stuff?
17:14:27 <boris-42> stannie with having decorator that cleans only specified services?)
17:14:42 <hughsaunders> less cleaning, less waiting
17:14:47 <boris-42> hughsaunders yep
17:14:55 <boris-42> msdubov btw yes I am against
17:15:02 <boris-42> msdubov don't touch stress test
17:15:15 <msdubov> boris-42 why&
17:15:16 <boris-42> msdubov until you finish your patch with refactoring of utils
17:15:18 <msdubov> why?
17:15:22 <msdubov> boris-42 sure
17:15:32 <boris-42> msdubov or let Kun make it..
17:15:39 <boris-42> msdubov or finish it by self..
17:16:26 <msdubov> boris-42 I suppose we'll continue the work with Kun on the processing utils after we merge https://review.openstack.org/#/c/79970/
17:16:54 <boris-42> msdubov it is not related...
17:17:11 <boris-42> msdubov to your work at all
17:17:24 <boris-42> msdubov so pls concentrate on first patch with utils
17:17:30 <boris-42> msdubov cause it blocks everything
17:17:43 <boris-42> msdubov "stress" runner doesn't block anything
17:18:09 <msdubov> boris-42 ok
17:19:32 <boris-42> hughsaunders are you working on passing context to scenario?)
17:19:42 <hughsaunders> boris-42: yep, but slowly
17:19:54 <hughsaunders> needs big rebase
17:20:14 <boris-42> hughsaunders hmm I didn't touch that part too much?)
17:22:25 <boris-42> so okay let move
17:22:29 <boris-42> let's move
17:22:39 <boris-42> So I am going to finish work (by the end of this wekk)
17:22:41 <boris-42> week*
17:22:59 <boris-42> and build ContextManger that will handle what context to run and what context are required by benchmark
17:23:15 <boris-42> this will finish work on BenchmarkStuff
17:23:31 <boris-42> So I will concentrate on other parts of projects
17:23:38 <boris-42> E.g. tempest & deploy stuff
17:23:54 <boris-42> And thoughts about benchmarking in VMs
17:24:19 <boris-42> hughsaunders msdubov  stannie btw does anybody have any questions?)
17:24:29 <boris-42> I mean seems like it's clear what and why we are doing=)
17:24:55 <hughsaunders> seems clear to me
17:24:57 <boris-42> seems like Rally team works even without meetings =(
17:25:05 <boris-42> or probably I should be happy=)
17:25:19 <boris-42> # Tempest stuff
17:25:24 <msdubov> Yep, there is actually the guiding doc
17:25:28 <msdubov> so it's clear
17:25:28 <boris-42> #topic Tempest stuff
17:25:48 <boris-42> 1. We should improve config generation (to make it works in any case0
17:26:10 <boris-42> I think that somebody should start playing with it… and catching all mistakes in configs
17:26:51 <boris-42> 2. We should create benchmark scenario that will run tempest scenario
17:27:02 <boris-42> Seems like this work is quite close to be finished
17:27:22 <boris-42> we will use "verification" stuff to make proper configuration for tempest
17:27:34 <boris-42> and we will have TempestContext
17:27:54 <boris-42> that will "activate" virtualenv with all stuff from tepest and run tempest unit test
17:28:06 <boris-42> and in cleanup it will "deactivate"
17:28:26 <boris-42> So seems like it's clear here as well hughsaunders penguinRaider stannie msdubov ?
17:29:00 <msdubov> point 1. : Is it clear who will handle this "catching mistakes" stuff?
17:29:15 <msdubov> really lots of work as we discussed it today
17:29:27 <boris-42> msdubov probably you?)
17:29:52 <msdubov> boris-42 Certainly not this week then...
17:30:01 <boris-42> msdubov it's okay
17:30:02 <penguinRaider> msdubov, boris-42 I can do that maybe ?
17:30:09 <boris-42> penguinRaider yep sure
17:30:10 <msdubov> boris-42 Actually my thought was that this should be done by 2-3 people
17:30:17 <msdubov> so we could do it together
17:30:32 <boris-42> actually Olga is already working on this stuff
17:30:37 <boris-42> but she need help
17:30:51 <boris-42> cause task takes a lot of time and it's quite big
17:31:16 <amaretskiy> just to note, i am ready for some kind of work
17:31:25 <boris-42> amaretskiy okay nice
17:31:40 <boris-42> amaretskiy penguinRaider let's speak today and coordinate this work with Olga
17:31:48 <boris-42> tomorrow*
17:31:51 <amaretskiy> ok
17:32:18 <penguinRaider> sure
17:33:01 <boris-42> # Deploy stuff
17:33:07 <boris-42> #topic Deploy stuff
17:33:19 <boris-42> hughsaunders penguinRaider amaretskiy msdubov  guys we should help rediskin
17:33:34 <boris-42> to finish at least multi host engine
17:33:50 <msdubov> boris-42 Do you mean code reviews?
17:33:56 <msdubov> or other kind of help?
17:34:02 <boris-42> msdubov code reviews and testing patches
17:34:26 <msdubov> hard to do code review without testing patches in case of rediskin =)
17:34:39 <hughsaunders> yep, I removed -1 today, but need to retest
17:34:54 <boris-42> msdubov lol
17:35:33 <boris-42> #topic Free discussion
17:36:05 <boris-42> msdubov hughsaunders  penguinRaider amaretskiy does anybody have any questions about Rally/RoadMap/any stuff...
17:36:06 <boris-42> ?
17:36:23 <amaretskiy> i don't
17:36:28 <penguinRaider> boris-42, hughsaunders msdubov I need help regarding my application for gsoc
17:36:35 <hughsaunders> penguinRaider: what needs doing?
17:36:46 <penguinRaider> I need a unifying theme for the proposal
17:36:48 <boris-42> penguinRaider yep you should share with them
17:36:52 <tzabal> do you have any proposed changes for the output of the tasks?
17:37:05 <boris-42> tzabal hi!
17:37:08 <boris-42> tzabal ?)
17:37:11 <penguinRaider> hi tzabal :-)
17:37:19 <tzabal> hello :)
17:37:31 <boris-42> tzabal out of task you mean CLI or Graphics?)
17:37:47 <hughsaunders> penguinRaider: is your proposal available somewhere to read?
17:38:07 <penguinRaider> hughsaunders, yeah I made a rough first draft will share with you
17:38:09 <tzabal> i was thinking about putting the average, min, max results that we have in CLI, in the report of plot2htm
17:38:57 <mwagner_lap> is there a way to dump the actual openstack calls rally is making ?
17:39:24 <boris-42> mwagner_lap all calls?
17:39:34 <boris-42> mwagner_lap all API calls you mean?)
17:39:50 <boris-42> mwagner_lap actually if you run it with -d you will get all calls..
17:39:59 <mwagner_lap> ok cool
17:40:18 <boris-42> mwagner_lap it will be just debug mode (and all http request will be shown with all details)
17:40:36 <boris-42> tzabal so that is the work that started msdubov
17:40:48 <boris-42> tzabal but he didn't finished yet first patch
17:41:09 <boris-42> tzabal after he finish work you may share this task with Kun
17:41:20 <boris-42> msdubov ^ pls help with organizing work on this
17:41:23 <tzabal> boris-42 ok
17:41:35 <boris-42> mwagner_lap so actually I have one idea
17:41:46 <msdubov> boris-42 ok
17:41:46 <mwagner_lap> just one :)
17:41:48 <boris-42> mwagner_lap to collect more data form our benchmarking stuff
17:41:50 <boris-42> mwagner_lap =)
17:41:55 <boris-42> mwagner_lap one for you=)
17:41:59 <boris-42> mwagner_lap at this second=)
17:42:08 <tzabal> boris-42 do you have any proposed change for the plot2html stuff?
17:42:18 <boris-42> tzabal nope
17:42:24 * mwagner_lap waits anxiously
17:42:26 <boris-42> tzabal btw I have idea what you may do
17:42:35 <boris-42> tzabal until you are waiting for msdubov
17:43:01 <boris-42> tzabal draw histogram (with possibility on Y and duration on X)
17:43:05 <boris-42> tzabal interested ?
17:43:11 <tzabal> boris-42 yeap
17:43:22 <boris-42> tzabal okay I will share with details after meeting
17:43:26 <boris-42> mwagner_lap so
17:43:39 <boris-42> mwagner_lap actually as you see we are doing active pooling of resource
17:43:51 <boris-42> mwagner_lap e.g. checking status of VM until it becomes "Acitve"
17:43:54 <boris-42> mwagner_lap yep?
17:44:51 <mwagner_lap> boris-42, actually trying to figure out how to do that now
17:45:02 <boris-42> mwagner_lap it's already done..
17:45:10 <mwagner_lap> tryting to figure out how you bypass the instance database
17:45:30 <boris-42> mwagner_lap not sure that understand queiston=)
17:45:41 <boris-42> mwagner_lap First step send request to create VM
17:45:56 <boris-42> mwagner_lap second step in while do get_vm_by_uuid
17:46:14 <boris-42> mwagner_lap if it is Active stop while, if it is in Error state raise Exception
17:47:06 <mwagner_lap> as I mentioned earlier this morning, if i run the "boot" task, vms get created and are left running but they dont show up via nova list and they are not in the instance database
17:47:26 <mwagner_lap> which doesn't delete them from the db
17:47:50 <mwagner_lap> they show up in a ps -ef | grep qemu on the hypervisor but in virsh list --all
17:48:13 <mwagner_lap> so procs are running but no record of them from the openstack or libvirt perspective
17:49:24 <boris-42> mwagner_lap so seems like you didn't get my response
17:49:49 <boris-42> mwagner_lap the reason why you didn't see them is that you try to list admin's VMs
17:49:58 <boris-42> mwagner_lap and Rally creates VMs from temporary users
17:52:55 <mwagner_lap> boris-42, i did get that but shouldn't all the vms show up in the "systems" instance table ?
17:53:24 <boris-42> mwagner_lap "systems"?
17:53:34 <mwagner_lap> from an openstack pov there should be an accounting of the resources used
17:53:35 <boris-42> mwagner_lap they should be showed in "instances" table
17:54:10 <mwagner_lap> they are not showing up in mine, but they are running on the hypervisor
17:56:48 <boris-42> mwagner_lap hmm strange
17:56:58 <boris-42> mwagner_lap they should be all in DB
17:57:10 <boris-42> mwagner_lap and if you are looking in Horizon
17:57:27 <boris-42> mwagner_lap they should be listed in "instances"
17:57:30 <mwagner_lap> horizon does not show them and 'virsh list' doesn't list them either
17:57:33 <boris-42> mwagner_lap at admin page
17:57:38 <mwagner_lap> virsh list on the hypervisor
17:57:56 <boris-42> mwagner_lap okay lets disccuss this in rally chat
17:57:59 <mwagner_lap> bogdando, right not even with the admin page
17:58:04 <mwagner_lap> ok
17:58:08 <boris-42> #endmeeting