16:01:32 <jd___> #startmeeting
16:01:32 <jd___> #meetingname ceilometer
16:01:32 <jd___> #link https://lists.launchpad.net/openstack/msg13745.html
16:01:33 <openstack> Meeting started Thu Jun 28 16:01:32 2012 UTC.  The chair is jd___. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:35 <openstack> The meeting name has been set to 'ceilometer'
16:02:10 <jd___> #topic dachary: talk to Dragon about SystemData / ceilometer and try to create cooperation
16:02:39 <jd___> not sure he's here actually
16:02:45 <jd___> #topic dhellmann: talk to Quantum devs about integration with ceilometer
16:02:58 <jd___> I saw you did that dhellmann
16:02:58 <dhellmann> I sent email to the list to start that discussion
16:03:12 <dhellmann> https://lists.launchpad.net/openstack/msg13753.html
16:04:29 <jd___> nothing else to add from anybody?
16:05:00 <flacoste> no
16:05:00 <dhellmann> it's going to take a bunch of work to integrate with quantum, but we might be able to get most of what we want from asking libvirt about traffic
16:05:16 <dhellmann> we won't get the create/delete events, but they don't have notification for those yet anyway
16:05:47 <jd___> maybe but libvirt isn't going to take us far neither
16:05:54 * dhellmann apologizes for the delayed response. internet is down at the office, so I'm at a restaurant
16:06:25 <dhellmann> we may be able to get traffic stats, but I don't know how useful they will be. they won't differentiate between "public" and "private" traffic
16:06:57 <jd___> indeed
16:06:58 <DanD_> we have to use IP addresses to differentiate billable vs. non-billable traffic
16:07:15 <DanD_> a bit of a hack but it works
16:07:23 <jd___> ok, I'll move on next topic since this not related directly to this action anyway
16:07:38 <dhellmann> right, at dreamhost we're going to have our firewall VMs collect the traffic data and publish metering events
16:07:50 <jd___> DanD_: differentiate is doable, capture/count for each is another problem :)
16:08:04 <jd___> #topic dhellmann and nijaba to join the cinder meeting wed at 16UTC
16:08:18 <jd___> I think nijaba is away, so dhellmann again :)
16:08:25 <dhellmann> I was late but nijaba was there the whole time
16:08:39 <dhellmann> we have a similar situation: they don't have RPC or notifications, yet
16:08:49 <dhellmann> so we discussed creating an admin API that ceilometer could poll
16:09:11 <dhellmann> I think that's going to mean we want another type of agent daemon that runs somewhere other than the compute nodes (probably on the same host as the collector)
16:10:00 <dhellmann> that sort of ties into the message I sent to the mailing list this morning, but we can discuss that at the end if we have time
16:10:42 <dhellmann> that's all I have
16:10:46 <jd___> ok fine!
16:11:05 <jd___> #topic jd_: check that openstack-ci runs the tests for Essex
16:11:28 <jd___> I did check that with dhellmann
16:11:37 <dhellmann> are those tests passing now?
16:11:42 <jd___> yes
16:11:51 <jd___> and AFAIK they are running automagically on review requests
16:11:59 <dhellmann> ok, we should turn the gates back on if they pass
16:12:37 <jd___> I think it's on, isn't it?
16:12:57 <dhellmann> I turned them on for patch uploads, but not for gating after approval
16:14:17 <jd___> ok
16:14:38 <dhellmann> they are listed in separate places in the zuul config file
16:15:05 <dhellmann> we can talk about that by email, though
16:15:10 <dhellmann> what else do we have today?
16:15:21 <jd___> #topic jd_: add some tests on daemon launching
16:15:31 <dhellmann> ah, yeah, that would be very useful
16:15:39 <jd___> did that to some extents, now it works fine on Essex
16:15:53 <jd___> I think it used to fail before
16:16:13 <jd___> #topic nijaba: to prime the doc on plugins
16:16:21 <jd___> well, nijaba did it we saw it
16:16:23 * dhellmann thinks he has about 10 minutes before the restaurant asks him to clear out and make room for the lunch crowd
16:16:36 <dhellmann> yes, and we had some updates from annegentle, too
16:16:42 <jd___> should be
16:16:43 <jd___> https://github.com/stackforge/ceilometer/commit/1b7b651d860a8fe38d523f67cb229302159d0872
16:17:00 <jd___> dhellmann: ok I'll hurry :)
16:17:08 <jd___> #topic nijaba: to question best method to authenticate admin client calls from ceilometer to cinder (and other projects)
16:17:09 <jd___> 
16:17:15 <jd___> something to say about that dhellmann ?
16:17:18 <flacoste> i think that's also done
16:17:37 <jd___> flacoste: ok, thanks :)
16:17:37 <dhellmann> we did discuss it, but I'm not sure it was resolved
16:17:39 <dhellmann> it's in process
16:17:40 <flacoste> this was the suggestion to have some API calls restricted to the local ip
16:17:45 <flacoste> right
16:17:59 <flacoste> i don't recall a reply from cinder there
16:18:02 <dhellmann> nova and some of the other projects have "admin" APIs, so we should try to mimic what they did
16:18:07 <flacoste> did you discuss it in your meeting?
16:18:26 <DanD_> you might want to consider that the metering system will potentially be distributed to scale, this will require remote communication
16:18:51 <DanD_> ours already consists of multiple systems
16:18:53 <dhellmann> yeah, we won't want to restrict to local IP or force any particular configuration
16:18:58 <dhellmann> although if it's just an agent, that's not as bad
16:19:22 <dhellmann> flacoste, we did discuss it but I don't think we resolved it
16:19:35 <dhellmann> there were a couple of ideas and we basically said "yes, we can figure this out"
16:19:49 <jd___> which is probably true :)
16:20:39 <dhellmann> I hope so :-)
16:21:08 <dhellmann> anything else on the agenda for today?
16:21:18 <jd___> #topic Discuss and vote the incubation application proposal to be submitted to PPB
16:21:37 <jd___> not sure we should vote this today, but we can discuss
16:21:47 <dhellmann> I haven't had a chance to read that yet. Is nijaba here? I wouldn't want to vote without him and Loic present.
16:21:58 <dhellmann> jd___ do you have the link to that handy?
16:22:05 <jd___> http://wiki.openstack.org/Projects/IncubatorApplicationCeilometer
16:22:33 <jd___> no I think they're both away so we should reschedule the vote
16:22:46 <flacoste> fwiw, that looked fine to me
16:22:52 <dhellmann> and probably the discussion, since if we have questions there won't be anyone to answer :-)
16:23:02 <jd___> indeed
16:23:34 <jd___> #topic Open discussion
16:23:58 <jd___> (if dhellmann didn't start cooking yet)
16:24:10 <dhellmann> I'm still here
16:24:38 <jd___> any point we want to discuss now rather than by email?
16:24:39 <dhellmann> the only thing I had was the API change I proposed on the list. I saw jd___'s response but have been offline for 3-4 hours so couldn't reply
16:24:53 <dhellmann> email is probably better for that discussion anyway
16:25:06 <jd___> yeah, that's fine with me too
16:25:17 <dhellmann> ok
16:25:41 <dhellmann> oh, I've made some good progress with the mongodb driver. I should have something to put up for review early next week
16:25:54 <jd___> dhellmann: great, your finally went for mongo?
16:25:59 <jd___> -r
16:26:01 <dhellmann> I'm working out some tricky queries right now
16:26:27 <dhellmann> yeah, the ops team looked at the calculator nijaba put together and decided it wasn't so much data that they were worried about mongo's performance
16:26:39 <jd___> ok
16:26:40 <dhellmann> at least not for our beta
16:26:57 <dhellmann> if they end up changing their mind, I will have to write another driver :-)
16:27:06 <jd___> :-)
16:27:35 <jd___> anything else?
16:27:47 <dhellmann> one thing that came up is the API we have discussed before may need to expand so that when you ask for sums of values for a resource, you get multiple answers, one per unique metadata set.
16:27:53 <dhellmann> I will write more about that on the mailing list
16:28:22 <jd___> dhellmann: ok
16:29:20 <dhellmann> that's all I had for today
16:29:37 <jd___> I don't have more for now neither
16:29:46 <jd___> #endmeeting