14:02:45 <boris-42> #startmeeting Rally
14:02:50 <openstack> Meeting started Mon Oct 12 14:02:45 2015 UTC and is due to finish in 60 minutes.  The chair is boris-42. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:51 <e0ne> hi
14:02:52 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:53 <boris-42> ping everybody
14:02:55 <openstack> The meeting name has been set to 'rally'
14:03:05 <amaretskiy> hi
14:03:07 <lcazares> hello
14:03:23 <ylobankov> hi
14:03:51 <boris-42> redixin: oanufriev_ andreykurilin ping
14:04:04 <andreykurilin> \o/
14:04:28 <boris-42> let's wait few more seconds
14:04:41 <redixin> dzien dobry
14:04:55 <rvasilets> hi
14:05:14 <boris-42> mbonell_: hi
14:05:22 <oanufriev_> hi all
14:05:34 <boris-42> okay GO GO GO GO
14:05:47 <boris-42> topic [mbonell] Blueprint discussion. New command to install Tempest's plugins in rally verify  - https://blueprints.launchpad.net/rally/+spec/install-tempest-plugins
14:06:03 <boris-42> mbonell_: ^ so
14:06:31 <rvasilets> #link https://blueprints.launchpad.net/rally/+spec/install-tempest-plugins
14:08:36 <boris-42> mbonell: So I believe that this is nice feature
14:08:48 <boris-42> mbonell: maybe we should call the command rally verify install-plugin
14:08:56 <mbonell> sorry problems with my connection
14:08:59 <boris-42> _matthias_: instead of just pluign*
14:09:09 <boris-42> _matthias_: sorry*
14:09:13 <mbonell> ok, yes sure! is a better name
14:09:14 <boris-42> mbonell: ^
14:09:28 <boris-42> mbonell: we already have install command so it will be unfied
14:09:40 <boris-42> e.g. you can install tempest or it's plugins
14:09:40 <mbonell> yes
14:09:48 <boris-42> mbonell: otherwise it looks good to me
14:09:56 <boris-42> ylobankov: andreykurilin ^ what do you think guys?
14:10:01 <mbonell> tranks
14:10:05 <mbonell> thanks*
14:10:33 <andreykurilin> ` rally verify install-plugin` sounds good to me
14:10:57 <ylobankov> I like `rally verify install-plugin` :)
14:11:05 <mbonell> :)
14:11:10 <lcazares> me too
14:11:28 <boris-42> Okay so let's just do the quick fix in BP and get it approved
14:11:46 <mbonell> ok!
14:12:01 <boris-42> mbonell: it's approved https://blueprints.launchpad.net/rally/+spec/install-tempest-plugins
14:12:10 <boris-42> let's move to next topic ?
14:12:22 <mbonell> and is my first BP, so what are the next steps? Just star working on it?
14:12:43 <boris-42> mbonell: actually you could just make a code even without BP =)
14:13:00 <boris-42> mbonell: so yep just make some code and do the git review
14:13:12 <mbonell> excellent!
14:13:51 <boris-42> mbonell: thanks for working on this=)
14:13:57 <boris-42> #topic  [lcazares] Blueprint discussion. Add optional parameters at rally container creation to auto-register an Openstack deployment.
14:14:03 <boris-42> lcazares: hi there
14:14:07 <lcazares> yes
14:14:08 <boris-42> lcazares: so what is the case for this?
14:14:58 <lcazares> just to have a working container after it is up... I mean with a deployment already registered
14:15:57 <lcazares> that way if I need a quick test I can create/ destroy the container with less commands...
14:16:19 <yfried> hi
14:16:43 <boris-42> lcazares: btw does docker image accept arguments?
14:16:49 <redixin> lcazares, there is one problem
14:16:51 <boris-42> lcazares: didn't see that ..
14:16:55 <redixin> boris-42, it accept
14:16:57 <redixin> but
14:17:00 <boris-42> redixin: hm nice
14:17:14 <boris-42> yfried: hi
14:17:38 <redixin> lcazares, where rally database will be stored?
14:17:47 <boris-42> redixin: so I have answer on this
14:17:55 <boris-42> redixin: we should refactor our Docker image
14:18:03 <boris-42> redixin: to use sqlite inside it
14:18:18 <boris-42> redixin: and provide instruction how to make persistance storage
14:18:32 <redixin> ok. then we have other problem
14:18:40 <boris-42> redixin: what is the other problem?)
14:19:03 <redixin> changes will not be saved after we exited container
14:19:14 <redixin> and changes in rally.db will be lost
14:19:25 <boris-42> redixin: it's not the problem
14:19:37 <boris-42> redixin: as I said we will make the insturction how to make persistance storage
14:20:05 <redixin> we already have docs how to make persistent storage
14:20:07 <stpierre> we already have that
14:20:22 <boris-42> redixin: stpierre I know
14:20:32 <redixin> so what will be changed?
14:20:44 <boris-42> redixin: rally-manage db recreate called in container
14:20:50 <redixin> we add "rally-manage db recreate" to Dockerfile?
14:20:55 <boris-42> redixin: Yep
14:21:11 <stpierre> you can't map volumes during container build
14:21:13 <stpierre> so that won't work
14:21:28 <redixin> and if we use external volume, it will "overload" default storage, right?
14:22:27 <stpierre> yes
14:23:01 <boris-42> redixin: stpierre hm there is no way to create file in docker container?
14:23:06 <redixin> so we have two options: built-in empty database (or not empty). this changes to this database are lost after exiting container (unless we commit them)
14:23:08 <stpierre> you can create a file in the container
14:23:08 <boris-42> that will contain DB?
14:23:22 <redixin> and second option is what we have now
14:23:27 <stpierre> but if we create the db in the Dockerfile, then anyone using persistent storage will have to recreate it; and anyone not using persistent storage will lose any data they store in the database between invocations of the container
14:24:21 <boris-42> stpierre: those who are using persistance storage are already quite f**ed up by installation tutorial
14:24:33 <stpierre> and this still doesn't get us closer to the ultimate desired feature, to auto-register a deployment, since you can't pass arguments to 'docker build'
14:24:41 <boris-42> stpierre: and in most case people do not need results
14:24:47 <boris-42> stpierre: after termitnation of container
14:25:23 <stpierre> it's not just results, it's also things like deployments. without persistent storage you have to recreate your deployment(s) every time you run the container
14:25:24 <boris-42> lcazares: btw you can have persistance storage
14:25:47 <boris-42> stpierre: but why?
14:25:49 <stpierre> yeah, the solution to this feature request really seems to be persistent storage, which is already possible and documented
14:25:56 <boris-42> stpierre: you can register it during the build or not?
14:25:59 <stpierre> how?
14:26:05 <stpierre> you'd have to explicitly write it in to the dockerfile
14:26:13 <stpierre> because you can't pass arguments to the build
14:26:19 <boris-42> stpierre: ah
14:26:35 <boris-42> stpierre: in any case I was asked few times already
14:26:49 <boris-42> stpierre: to make docker image work without magic and persistance storage
14:26:58 <boris-42> stpierre: just to work out of box
14:27:05 <redixin> this is nature of docker -_-
14:27:27 <stpierre> creating the database in the Dockerfile mostly does that, but in a way that I think is fairly unintuitive.
14:27:55 <stpierre> but it doesn't make things any worse for people who want to run with persistent storage, so it's not all bad
14:28:23 <boris-42> redixin: nature of docker is pain?
14:28:42 <redixin> nature of docker is "be useless"
14:28:53 <boris-42> redixin: stpierre so let's just run rally-manage db recreate (if it does not use persistance storage)
14:29:00 <boris-42> redixin: stpierre can we check that on fly?
14:29:05 <stpierre> no
14:29:14 <boris-42> stpierre: we can
14:29:32 <boris-42> stpierre: we can check whatever DB exist or not
14:29:33 <stpierre> no, you can't. volume mapping is determined at container run time, not build time.
14:29:50 <redixin> seeing message "run rally-manage db recreate" is enough as for me
14:30:04 <stpierre> we can just create the database during the build; if someone uses persistent storage, they'll mount a volume over it
14:30:14 <boris-42> stpierre: and it will work ?
14:30:21 <stpierre> fsvo "work"
14:30:22 <boris-42> stpierre: if so let's do it
14:30:40 <boris-42> stpierre: I mean it won't purge each time persistance storage
14:30:48 <stpierre> correct
14:30:54 <redixin> so it all about just adding "rally-manage db recreate"?
14:31:07 <redixin> also user can commit changes after each run
14:31:23 <boris-42> redixin: yep
14:31:36 <boris-42> redixin: so let's add this one line
14:32:01 <boris-42> lcazares: and in your case you should just create once persistance storage like in docs
14:32:09 <boris-42> is describe
14:32:13 <redixin> stpierre, is it possible to access env during build?
14:32:30 <redixin> i mean if we run '. openrc admin admin' and run 'docker build'
14:32:36 <stpierre> nope
14:32:39 <redixin> sad
14:32:46 <boris-42> redixin: mreeeee pain
14:33:54 <redixin> so we only can add tutorial 'how to create deploymet in docker images having only . openrc admin admin'
14:33:59 <redixin> or something similar
14:35:34 <boris-42> redixin: hehe maybe
14:35:49 <boris-42> redixin: stpierre https://review.openstack.org/233665
14:35:55 <boris-42> okay let's move to next topic
14:36:04 <lcazares> just a quick comment
14:36:06 <boris-42> #topic [amaretskiy] What are the most critical bugs and features (required) in HTML reports?
14:36:13 <boris-42> lcazares: sure
14:36:22 <boris-42> lcazares: so?
14:36:23 <lcazares> the env could be passed at docker run instead of docker build
14:36:32 <amaretskiy> as already discussed with boris-42, there are 2 major tasks for html reports
14:36:41 <boris-42> amaretskiy: wait please
14:36:44 <amaretskiy> ok
14:37:14 <boris-42> lcazares: but it just opens console
14:37:18 <lcazares> so the idea would be to add the deployment data when doing the docker run
14:37:41 <redixin> lcazares, it is possible already
14:37:59 <lcazares> ok, I'll take a look
14:38:04 <lcazares> thanks guys
14:38:06 <boris-42> lcazares: you can pass whole DB of Rally to docker
14:38:06 <redixin> lcazares, we can do anything and commit changes, or we can do anything and use persistan storage
14:38:21 <boris-42> lcazares: from your disk
14:38:36 <boris-42> lcazares: or commit after you make proper changes like redixin is saying
14:39:08 <redixin> boris-42, it opens console by default. you can pass any command and arguments to 'docker run'
14:39:34 <lcazares> ok :)
14:40:07 <boris-42> okay
14:40:07 <redixin> docker run rallyforge/rally rally-manage db recreate # it will create db. and there is no need to commit if we use persistent storage
14:40:09 <boris-42> amaretskiy: your turn
14:41:09 <amaretskiy> it looks clear with html reports - 2 major tasks are 1) rework reports UI for new task format and 2) rework scenarios output
14:41:29 <amaretskiy> does any have anything to add?
14:41:34 <boris-42> amaretskiy: scenario output include the stuff related in scenario base
14:41:57 <boris-42> amaretskiy: I mean it should be refacotre do extend the schema of output
14:42:09 <amaretskiy> yes, I will add a new method to Scenario class
14:42:18 <boris-42> amaretskiy: I believe you should write spec before doing anything
14:42:28 <amaretskiy> boris-42: ok
14:42:31 <amaretskiy> will do
14:42:33 <boris-42> amaretskiy: because I am not sure that somebody is understanding about what you talking
14:42:37 <boris-42> amaretskiy: except me=)
14:42:42 <amaretskiy> okay
14:43:02 <amaretskiy> spec only for for reworking scenarios output charts?
14:43:10 <boris-42> amaretskiy: yep
14:43:19 <boris-42> amaretskiy: and not only charts
14:43:20 <amaretskiy> for UI for new task format?
14:43:24 <boris-42> amaretskiy: whole refactoring
14:43:34 <boris-42> amaretskiy: nope I don't care about that part *
14:43:39 <amaretskiy> okay
14:43:47 <boris-42> amaretskiy: just make it pretty
14:44:11 <boris-42> okay let's do the next topic
14:44:14 <boris-42> #topic  [redixin] Docker pull downloads all tags by default
14:44:18 <amaretskiy> UI for new task format requires a lot of changes in template/JS
14:44:20 <amaretskiy> ok
14:46:08 <boris-42> redixin: are you arround?
14:46:17 <redixin> yep
14:46:26 <boris-42> redixin: so any comments ideas I don't know
14:46:39 <redixin> so amaretskiy confirmed that docker pull rallyforge/rally:latest download only latest
14:46:47 <boris-42> redixin: yep
14:46:57 <boris-42> redixin: and?
14:47:17 <redixin> I was thinking do not use tags for releases
14:47:52 <boris-42> redixin: I do not understand you =(
14:47:56 <redixin> but use different images instead. like 'docker pull rallyforge/rally-0.1.1'
14:48:13 <boris-42> redixin: that may be a good idea
14:48:29 <boris-42> and have rally-latest image
14:48:37 <redixin> or just 'rally'
14:48:40 <boris-42> yep
14:48:57 <boris-42> because pepole usually are doing docker pull rallyforge/rally
14:49:11 <boris-42> and then it starts ........ using all your HD
14:49:18 <boris-42> or SSD =)
14:49:23 <redixin> and all your internets
14:50:29 <redixin> currently it takes 1.1 gigabytes
14:50:39 <redixin> what will be later?
14:50:50 <redixin> when someone run 'docker pull rallyforge/rally'
14:51:03 <redixin> and download all 24 releases
14:51:10 <redixin> 4G
14:51:18 <boris-42> redixin: ya that sux
14:51:34 <stpierre> can't we delete old tags and only maintain the last 2 or 3 or N releases?
14:51:34 <boris-42> redixin: so I like your idea
14:51:44 <boris-42> stpierre: nope
14:51:49 <stpierre> :(
14:51:50 <boris-42> stpierre: people is using old releases
14:51:59 <boris-42> stpierre: I know people that are using rally0.0.3
14:52:09 <stpierre> but that would be included in the last 2 releases
14:54:01 <andreykurilin> boris-42: `people is using old releases`. people uses old releases because they are supported
14:54:16 <boris-42> andreykurilin: it is not supported LOL
14:54:20 <boris-42> andreykurilin: it just exist LOL
14:54:36 <boris-42> stpierre: andreykurilin so in any case even if we support like 5 releases
14:54:52 <boris-42> it will be more then 1 GB to download each time
14:55:01 <boris-42> which is really bad
14:55:44 <stpierre> meh, docker eats disk like crazy. anyone running docker needs to be willing to deal with that, IMO. we can limit how much disk it eats, but that's pretty much what docker does best.
14:56:27 <boris-42> stpierre: so why not just having X images
14:56:56 <stpierre> afaik that's not The Docker Way™
14:56:59 <boris-42> stpierre: we can actually have for each release branch like 0.1, 0.2, .... single image with bunch of tags
14:57:09 <boris-42> stpierre: but in rallyforge/rally there will be only one image
14:57:18 <boris-42> stpierre: latest master
14:57:21 <stpierre> so it'd be a little strange. not necessarily bad.
14:57:30 <boris-42> stpierre: and in rallyforge/rally-stable the latest release
14:58:04 <boris-42> stpierre: or we can have rally-all to put all the shit inside it
14:58:15 <redixin> we can do nothing for now, but back to this topic later :]
14:58:17 <stpierre> right, i understand the proposal, it's just not The Docker Way. that's precisely what tags are for. we're reinventing tags with image names to save disk space
14:58:30 <stpierre> which, like i said, isn't necessarily *bad*, just strange.
14:58:41 <boris-42> ok
14:58:49 <boris-42> I have some small update as well
14:58:55 <redixin> probably they change default behavior
14:59:04 <boris-42> #topic ceritification task
14:59:15 <boris-42> we are working on this https://github.com/openstack/rally/tree/master/certification/openstack
14:59:25 <boris-42> One single task that can do the properl load for YOUR cloud
14:59:39 <boris-42> so please join us in this effort *
14:59:46 <boris-42> #endmeeting