14:59:59 <sigmavirus> #startmeeting craton
15:00:00 <openstack> Meeting started Mon Jan  9 14:59:59 2017 UTC and is due to finish in 60 minutes.  The chair is sigmavirus. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:01 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:04 <openstack> The meeting name has been set to 'craton'
15:00:05 <openstack> jimbaker: Error: Can't start another meeting, one is in progress.  Use #endmeeting first.
15:00:08 <sigmavirus> #chair jimbaker
15:00:08 <openstack> Current chairs: jimbaker sigmavirus
15:00:51 <jimbaker> sigmavirus, that was pretty awesome synchronization
15:01:15 <sigmavirus> agreed
15:01:40 <jimbaker> and a good test of the meetbot system, one in fact will win :)
15:04:04 <sigmavirus> anyway
15:04:18 <jimbaker> yep, time to get started
15:05:17 <jimbaker> git-harry, palendae, sulo, hi
15:06:08 <git-harry> Hi
15:06:18 <jimbaker> sigmavirus, anything you would like to add to our usual standing agenda of what we plan to do this week?
15:07:27 <jimbaker> ok, sounds like it's a great plan :) but we can add more as we discuss
15:08:14 <sigmavirus> :)
15:08:22 * sigmavirus suspects sulo is here as sacharya
15:09:01 <jimbaker> sigmavirus, you spooked him
15:09:47 <jimbaker> so last week i made good progress on rbac, especially understanding how to merge the scoped role assignments with oslo.policy rules. it's pretty straightforward
15:11:08 <jimbaker> which means, i have a working demo script of what it should look like; and i have been able to figure out what oslo.policy actually does, despite the rather cryptic docs. mostly by reading keystone and its own codebase
15:12:05 <jimbaker> this week i plan to get that in as a spec for review. about time
15:12:54 <jimbaker> i'm also going to set up meetings with dusty simonyi to help us on this phase of requirements gathering
15:13:14 <jimbaker> lastly i hope to put together a roadmap draft
15:13:29 <jimbaker> so that's my week
15:14:26 <jimbaker> sigmavirus, what are things looking like for you? it seems we got the pagination spec in good shape for completion shortly, and some discussion with the api wg as well on that
15:16:47 <sigmavirus> I'm going to wrap up the specification and then work on the actual implementation
15:17:54 <jimbaker> sigmavirus, cool. i think it's just a matter of some copy editing, and the spec will be wrapped up for craton
15:18:52 <jimbaker> and thanks again for that spec work, it's really well done/well thought out
15:20:08 <jimbaker> in particular, capturing alternatives (and their counterargs) in the spec is something we should work on doing for future specs
15:20:35 <sigmavirus> That's why there's a section for it
15:20:36 <jimbaker> git-harry, anything you would like to report?
15:20:40 <jimbaker> sigmavirus, :)
15:21:18 <jimbaker> a good template, a good process, certainly builds a foundation for good work
15:21:37 <git-harry> jimbaker: I have some things up for review, other than that nothing is new.
15:22:24 <jimbaker> git-harry, sounds good, and we will take a look at those reviews. thanks for those!
15:24:42 <jimbaker> on friday, we had a good discussion with respect to instrumentation for distributed tracing. it's not something that means so much now, given the simplicity of how inventory works today. it will be far more useful as we look at integration with other sources of truth (eg nova plugins with virtualized variables)
15:25:20 <jimbaker> it is not something i'm really working on this week, given current low pri, but just something we will want to start thinking about
15:26:57 <jimbaker> in particular, this means implementing http://docs.openstack.org/developer/osprofiler/ support
15:27:12 <sigmavirus> I've heard ... not great things about osprofiler
15:28:27 <jimbaker> sigmavirus, got it. and that's quite valid feedback. just want to understand better pros/cons, and possibly that there specific aspects that can make it work better or not
15:29:22 <jimbaker> in the meantime, we have this new distributed tracing functionality in aws, now under preview, https://aws.amazon.com/xray/
15:29:52 <jimbaker> which complements (and possibly builds on) related work in terms of cloudtrails
15:31:02 <jimbaker> sigmavirus, is it also possible that osprofiler's issues are more with respect to ceilometer, vs the api itself?
15:31:33 <sigmavirus> I believe it's more due to the fact that even when it's disabled it causes a performance hit
15:32:24 <sigmavirus> I don't have numbers for that though
15:32:25 <jimbaker> sigmavirus, got it. so maybe more of a question of appropriate granularity
15:34:00 <jimbaker> since osprofiler is new to me, i don't have a sense of how people use/abuse it. from the DTS perspective, it's more about handoffs between distributed components, and that's where i would see any usage for us making sense
15:34:40 <jimbaker> anyway, low pri for now
15:34:43 <sigmavirus> jimbaker: I don't think it has to do with abuse
15:35:16 <jimbaker> sigmavirus, cool
15:35:41 <sigmavirus> As i understand it, they're using osprofiler as documented
15:35:41 <jimbaker> sigmavirus, are people using it for finegrained, or coarsegrained?
15:36:31 <jimbaker> the other thing i have been considering is that we may want to organize keys with respect to some namespace system. right now we have a flat namespace for the keys themselves, although something like ansible can take that and apply whatever convention is desired
15:36:59 <sigmavirus> jimbaker: afaik, osprofiler has no distinction. it's on or off
15:37:06 <sigmavirus> and when it's off, it still hurts
15:37:25 <jimbaker> sigmavirus, ok. worth profiling then ...
15:37:26 <jimbaker> ;)
15:38:42 <jimbaker> so the context for namespaces: if we are storing secrets like ssh keys; or plugin configs; or what have you, we may want to distinguish from usage like ansible
15:40:20 <jimbaker> i don't think this needs to be more than a convention, and possibly future indexing support. so if convention it could be "ssh/id_rsa", "plugin/nova", etc - in this particular case, use / as the namespace separator
15:40:42 <jimbaker> just putting this out there to start thinking about a namespace spec
15:40:54 <jimbaker> Syed__, hi
15:41:37 <jimbaker> palendae, this could extend to separating out osa stuff as well
15:42:01 <Syed__> jimbaker: hi, sorry for delayed presence,, was unable to connect to my irc
15:42:16 <jimbaker> Syed__, we have been discussing our upcoming plans for this week
15:42:53 <jimbaker> in particular, i mentioned i got a working script going showing the bare minimum necessary for integrating what we do with scoped role assignments with oslo.policy rules
15:43:04 <palendae> jimbaker: I'm not sure how we'd use it because Ansible has no concept of namespaces
15:43:17 <jimbaker> palendae, that's actually my point :)
15:43:45 <jimbaker> i just want ansible to get what it needs from our integration, without getting the stuff it doesn't want
15:44:26 <Syed__> jimbaker: yeap i have been playing around with policy.json for setting up rules, can you put up the script as well in gist somewhere
15:45:01 <jimbaker> palendae, we will want to be filtering this out regardless. using some sort of namespace will hopefully simplify what is necessary. but definitely want your feedback!
15:45:23 <jimbaker> Syed__, yes, i will put it out shortly; and it will also be described in the spec
15:45:26 <palendae> Ok; well, anything querying craton can reassemble things as it wants on the client side
15:45:43 <Syed__> cool, thanks
15:45:49 <jimbaker> palendae, exactly
15:47:03 <jimbaker> Syed__, but in a nutshell, it's more about how to use oslo.policy than anything clever. it's really simple. but figuring it out took awhile, given rather cryptic docs
15:47:37 <jimbaker> anything else? that's all i have
15:49:21 <jimbaker> sigmavirus, ?
15:49:34 <sigmavirus> I've nothing much
15:50:01 <jimbaker> sigmavirus, feel free to mention it. but we have #craton of course to go into more details
15:50:33 <palendae> https://review.openstack.org/#/c/410342/ would be nice to get opinions from Craton devs
15:50:45 <palendae> I've gotten feedback from sigmavirus already and some from other OSA people, which I do need to get back to
15:51:30 <jimbaker> palendae, sounds good. i was re-reading the feedback this morning. i will see what i can add to the discussion
15:52:34 <jimbaker> ok, let's bring this meeting to a close
15:52:40 <jimbaker> thanks everyone!
15:52:50 <jimbaker> #endmeeting