17:12:50 <boris-42> #startmeeting Rally
17:12:50 <openstack> Meeting started Tue Jan 13 17:12:50 2015 UTC and is due to finish in 60 minutes.  The chair is boris-42. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:12:51 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:12:52 <msdubov_> boris-42, hi!
17:12:54 <openstack> The meeting name has been set to 'rally'
17:14:01 <boris-42> amaretskiy: ping
17:14:07 <amaretskiy> hi
17:14:10 <boris-42> paboldin: hi
17:14:17 <paboldin> hi all
17:14:32 <oanufriev> hi
17:16:38 <boris-42> okay let's tart
17:16:48 <boris-42> msdubov_: could you share details about your work on docs
17:17:01 <boris-42> #topic Improve Rally Docs
17:17:14 <msdubov_> boris-42, sure
17:17:22 <msdubov_> here is the patch https://review.openstack.org/120018
17:17:48 <msdubov_> here is the link to the current state of the docs in this patch http://docs-draft.openstack.org/18/120018/29/check/gate-rally-docs/fd04e63//doc/build/html/
17:18:23 <msdubov_> Compared to what is now there in ReadTheDocs, the documentation has been greatly restructured and somewhat simplified
17:18:44 <msdubov_> I've accomplished the 4-steps tutorial through different use cases in Rally
17:18:53 <msdubov_> ("Rally Step-by-Step")
17:18:59 <msdubov_> this is going to be expanded furthermore
17:19:30 <msdubov_> but by now I think it's time to show this documentation to our presales manager who was interested in it, collect feedback and finally merge this patch
17:19:38 <msdubov_> and then continue this work in future patches
17:19:47 <boris-42> msdubov_: I already send to him
17:20:00 <msdubov_> boris-42, Oh great I missed it
17:20:02 <boris-42> msdubov_: seems like he didn't have enough tim eto go through it
17:20:21 <boris-42> msdubov_: okay seems much better then now
17:20:25 <boris-42> before*
17:20:27 <boris-42> lol
17:20:29 <msdubov_> Also I've made a patch to global requirements to support the ReadTheDocs theme in Gates
17:20:34 <msdubov_> boris-42, Hope so!
17:20:35 <msdubov_> eom
17:21:53 <boris-42> msdubov_: I think section about plugisn
17:22:01 <boris-42> msdubov_: is a bit not completed
17:22:10 <msdubov_> boris-42, You want more samples?
17:22:11 <boris-42> msdubov_:  http://docs-draft.openstack.org/18/120018/29/check/gate-rally-docs/fd04e63//doc/build/html/plugins.html
17:22:32 <boris-42> msdubov_: there is only sample for scenario, no sla, no context, ...
17:23:05 <msdubov_> boris-42, Yep, but the mechanism seems to be the same for scenario/sla/context
17:23:12 <msdubov_> boris-42, Anyway I can add more examples
17:23:41 <boris-42> msdubov_: it is simlir when you know whole Rally
17:23:59 <msdubov_> boris-42, agree
17:24:03 <boris-42> msdubov_: and context are different from scenarios
17:24:08 <boris-42> msdubov_: absolutelly differnet
17:24:25 <boris-42> msdubov_: and I think we can add here subsections http://docs-draft.openstack.org/18/120018/29/check/gate-rally-docs/fd04e63//doc/build/html/gates.html
17:24:33 <boris-42> msdubov_: cause it's hard to read/understand
17:24:43 <boris-42> msdubov_: okay let's move to other topics)
17:25:04 <boris-42> #topic VMTask reactoring
17:25:08 <boris-42> amaretskiy: ping
17:25:13 <amaretskiy> hi
17:25:16 <boris-42> amaretskiy: hi there
17:25:22 <boris-42> amaretskiy: what is the status of patch?
17:25:43 <amaretskiy> I've done improvements for scenario boot-runcommand-delete
17:25:58 <amaretskiy> the patch is https://review.openstack.org/#/c/144337/
17:26:19 <amaretskiy> now we can run this scenario for neutron-based cluster
17:26:56 <amaretskiy> and most of networking setup is automated
17:27:06 <amaretskiy> so we can just use network context
17:27:13 <boris-42> amaretskiy: could you prove it ?
17:27:16 <amaretskiy> and left all networking parameters default
17:27:18 <boris-42> amaretskiy: and add VMTask here https://review.openstack.org/#/c/144337/9/rally-jobs/rally-neutron.yaml
17:27:56 <amaretskiy> yes, reasonable
17:28:01 <amaretskiy> that will be done
17:28:09 <amaretskiy> eom
17:29:51 <boris-42> amaretskiy: ok I will take a look at that patch tomorrow
17:29:57 <amaretskiy> ok
17:30:01 <boris-42> msdubov_: andreykurilin1 paboldin ^ please do the same=)
17:30:13 <paboldin> ack
17:30:31 <boris-42> #topic Benchmarking in VMs
17:30:39 <boris-42> paboldin: do you have something to say about it?)
17:30:48 <paboldin> not much yet
17:31:14 <paboldin> trying to decide what to use to launch hundreds of benchmarks on hundreds of VMs
17:31:22 <paboldin> writing a proposal
17:31:39 <paboldin> that is it
17:32:39 <boris-42> paboldin: oslo.messaging?
17:32:48 <boris-42> paboldin: or tinny REST web server
17:32:48 <paboldin> boris-42: will take a look, thanks
17:33:01 <boris-42> paboldin: I even don't now what is better)
17:33:02 <paboldin> boris-42: this will require us to connect to each VM
17:33:05 <paboldin> this is bad
17:33:11 <boris-42> paboldin: nope
17:33:11 <paboldin> i would prefer a bus
17:33:36 <boris-42> paboldin: HTTP request will scale well
17:34:10 <paboldin> we will see when the proposal is ready
17:34:31 <boris-42> paboldin: ok great
17:34:37 <boris-42> paboldin: could you share link of proposal?)
17:34:49 <paboldin> boris-42: when it will be at least in a first draft version
17:35:49 <boris-42> paboldin: ok good
17:35:56 <boris-42> #topic MOS is pain
17:36:15 <boris-42> amaretskiy: around?
17:36:22 <amaretskiy> yes
17:36:24 <boris-42> So I talk with FUEL guys
17:36:30 <boris-42> they said next
17:36:42 <boris-42> 1) they blocked access from fuel master node
17:36:50 <boris-42> because of security issue
17:37:04 <boris-42> 2) they blocked ineternal and managment url becuase of security
17:37:39 <boris-42> 3) They add back door in HA proxy (that enpoint with tricky PORT)
17:37:59 <boris-42> that is managment url (but in public network)
17:38:02 <boris-42> So.......
17:38:27 <boris-42> they add that backdoor because they have to have abiility to run OSTF that requires admin
17:38:46 <boris-42> amaretskiy: ^ so seems something very bad
17:38:55 <amaretskiy> agreed
17:39:11 <boris-42> amaretskiy: I think we should ask to allow access to managment url
17:39:16 <boris-42> from FUEL master node
17:39:31 <amaretskiy> why? we do not run rally on master node
17:39:36 <amaretskiy> that is a rare case
17:39:52 <amaretskiy> nobody (except me) does that
17:40:22 <amaretskiy> and it is very hard to install rally on this node
17:40:36 <boris-42> amaretskiy: nope
17:40:40 <boris-42> amaretskiy: now it is simple
17:40:47 <amaretskiy> docker ?
17:40:50 <boris-42> amaretskiy: cause we have docker
17:40:52 <boris-42> amaretskiy: yep
17:40:59 <boris-42> amaretskiy: and it is installed on fuel master node
17:41:05 <amaretskiy> okay
17:41:27 <boris-42> amaretskiy: so it will be quit easy for end users
17:41:33 <amaretskiy> but whay should not we just add a keystone oparameter "endpoint" into deployment config?
17:41:38 <amaretskiy> this parameter is commin
17:41:45 <amaretskiy> *common
17:44:00 <boris-42> amaretskiy: it is common for or clients
17:44:24 <boris-42> amaretskiy: in current implementation we are passing it only to keystone
17:44:30 <amaretskiy> that is just yet another keystone client parameter
17:44:39 <boris-42> amaretskiy: nope
17:44:56 <boris-42> amaretskiy: we should change managment url for all clients
17:45:28 <boris-42> amaretskiy: so for example live migrate benchmark (that requires admin) won't work
17:45:51 <boris-42> amaretskiy: and every clients has own endpoint
17:46:02 <boris-42> amaretskiy: so some kind of hell
17:49:00 <boris-42> ok
17:49:08 <boris-42> so I will talk to FUEL guys
17:49:11 <boris-42> and let's seee
17:49:21 <boris-42> #topic Fuel Task Templates
17:49:30 <boris-42> #topic Rally Task Templates
17:49:40 <boris-42> So it is the part of work that I am working on
17:49:53 <boris-42> It allows us to create parametrized jinja2 tempaltes
17:49:58 <boris-42> that generates tasks
17:50:16 <boris-42> wich allows to make independed from OpenStack clouds size and configratuion tasks
17:50:28 <boris-42> https://review.openstack.org/#/c/137716/
17:50:39 <boris-42> patch is here ^ and I hope to merge it soon
17:51:11 <boris-42> anybody has anything to say?)
17:52:08 <amaretskiy> i think we can specify template context data from stdin
17:52:34 <amaretskiy> example:  echo args.json | rally task start ...
17:52:41 <amaretskiy> s/echo/cat/
17:52:44 <boris-42> hm
17:52:56 <boris-42> rally task start ... --task-args-file
17:52:58 <boris-42> ?)
17:53:10 <amaretskiy> ok, let it be so
17:53:23 <amaretskiy> --task-args-file is ok
17:54:38 <boris-42> amaretskiy: it's just already implemented
17:54:49 <boris-42> amaretskiy: I mean in future we can extend it even to stdin
17:54:58 <amaretskiy> ok
17:55:01 <boris-42> amaretskiy: but for now I think it's better to merge patch, cause it's already too big=)
17:55:14 <boris-42> Okay we need to finish meeting soon=)
17:55:19 <boris-42> so see you guys
17:55:28 <boris-42> #endmeeting