17:02:09 <thinrichs> #startmeeting CongressTeamMeeting
17:02:09 <openstack> Meeting started Tue Oct  7 17:02:09 2014 UTC and is due to finish in 60 minutes.  The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:02:10 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:02:12 <openstack> The meeting name has been set to 'congressteammeeting'
17:02:17 <rajdeep> hi
17:02:19 <jasonsb> hi
17:03:01 <thinrichs> sarob and arosen1 actually put up an agenda this week
17:03:06 <thinrichs> https://wiki.openstack.org/wiki/Meetings/Congress
17:03:11 <arosen1> I can go first if you want.
17:03:17 <thinrichs> arosen1: ok
17:03:33 <arosen1> cool, So I've been  working on the CI stuff. We now have a job that is voting running a few basic functional tempest tests.
17:03:39 <thinrichs> #topic status
17:03:50 <arosen1> Right now we're lacking tempest tests for the datasource drivers (neutron, keystone, ceilometer) this would be awesome if someone wants to help out with this.
17:03:57 <arosen1> I added coverage for nova here which might be a good example to follow: https://review.openstack.org/#/c/125501/
17:04:29 <thinrichs> This is definitely a huge step forward!
17:04:45 <thinrichs> I'll volunteer to write a tempest test or two this week.
17:05:07 <arosen1> i think after we add this coverage we can also add a few scenario  tests to give us even more coverage testing rules over multiple tables etc
17:05:38 <arosen1> I've also written a datasource driver for glance which also includes functional tests: https://review.openstack.org/#/c/126377/ would be nice to get some reviews on this if anyone has cycles.
17:05:39 <thinrichs> Pretty soon tempest tests will be part of the expected tests we write for new features.
17:05:57 <arosen1> thinrichs: i think that would be a good requirement to add
17:06:25 <arosen1> that's all from me though unless there are any questions
17:06:25 <thinrichs> Great!  So we're seeing more people having fulfilled their datasource driver obligation.
17:06:51 <thinrichs> alexsyip: want to go next?
17:07:19 <alexsyip> I wrote the new datasource superclass, and I’m converting the keystone driver to use the new superclass (fixing bugs along the way).
17:08:03 <thinrichs> Good stuff!  Want to give people an idea of why the new superclass will make it easier to write new drivers?
17:08:41 <alexsyip> In the past, each datasource driver needed to convert the API reponse data from the data source to congress columns explicitly.
17:08:45 <rajdeep> arosen1 noticed a version v2 in the glance driver name
17:09:14 <arosen1> rajdeep:  right; glance has a v1 and v2 api. So v2 means it talks to the glance_v2 api
17:09:34 <rajdeep> that is true for nova too
17:09:40 * arosen1 that's why i did that.
17:09:43 <rajdeep> and most other apis
17:09:45 <alexsyip> The new superclass will allow the programmer to write a data descriptor.
17:10:01 <arosen1> rajdeep:  well i think we are only talking to nova's 1_1 api right now and not nova's v2
17:10:07 <alexsyip> To give an idea of what I”m talking about, take a look at this preliminary version: https://review.openstack.org/#/c/125845/1/congress/datasources/datasource_driver.py
17:10:08 <arosen1> the nova v2 is still a wip
17:10:20 <alexsyip> There’s a comment at the top describing the descriptor format.
17:10:30 <arosen1> alexsyip: sounds good, i'll try and also review this later today.
17:10:38 <alexsyip> It’s not ready for review.
17:10:46 <arosen1> okay i'll wait then :)
17:11:01 <alexsyip> But it would be valuable to look it over and see if the descriptors make sense.
17:11:08 <alexsyip> Just the comment part.
17:11:16 <alexsyip> Although it’s out of date..
17:11:29 <alexsyip> I’ll upload a newer version today.
17:12:14 <thinrichs> Sounds good!
17:12:42 <thinrichs> rajdeep: want to go next?
17:12:49 <rajdeep> sure
17:13:21 <rajdeep> i have started implementing horizon panel for datasource APIs
17:13:43 <rajdeep> will take some time to get it completed
17:14:07 <arosen1> rajdeep:  are you in sync with janet? have you seen this one: https://review.openstack.org/#/c/123912/ ?
17:14:22 <rajdeep> noticed that congress returns data source driver list for neutron even when neutron is disabled in devstack
17:14:56 <arosen1> rajdeep:  right, that happens because neutron is in the datasource.conf
17:15:09 <thinrichs> rajdeep: the status reflects whether Neutron is enabled in Congress.  Not whether it is enabled in devstack.
17:15:17 <arosen1> once datasource registration occurs via the api that won't be the case as someone will explicitly have to register it.
17:15:26 <thinrichs> Ideally we could do both.
17:15:30 <arosen1> thinrichs: we have a status now?
17:15:38 <thinrichs> But once we get the datasource statuses in place, we should get that info.
17:15:39 <rajdeep> ok makes sense but i got a little confused
17:15:47 <arosen1> ah yea..
17:15:53 <arosen1> nvm
17:16:18 <rajdeep> will we expose enabling drivers via horizon too?
17:16:30 <thinrichs> rajdeep: did you see the link arosen1 linked to?
17:16:41 <arosen1> rajdeep: i would think so.
17:16:42 <rajdeep> arosen1 i took the initial patch from janet and worked on top of it
17:16:51 <arosen1> rajdeep: great
17:17:09 <thinrichs> rajdeep: eventually we'll let people control drivers via the API and horizon.  But that's low priority right now.
17:17:16 <rajdeep> ok..
17:18:11 <rajdeep> that completes the status from my side
17:18:20 <thinrichs> sarob: you want to do your stuff?
17:19:01 <sarob> sooo, we almost have our space allocated for the summit
17:19:23 <rajdeep> another thing i forgot to mention
17:19:35 <sarob> go for it rajdeep
17:19:51 <rajdeep> horizon needs an id for each row else it is not able to render properly
17:20:09 <rajdeep> datasource row doesn't have an id
17:20:21 <rajdeep> janet also mentioned this in her comments
17:21:56 <thinrichs> Giving rows IDs is hard.
17:22:04 <thinrichs> Not all rows are stored explicitly—they are computed on the fly.
17:22:20 <thinrichs> So giving IDs doesn't always work, which means we don't give IDs to rows.
17:22:49 <thinrichs> Why can't we render rows without IDs?
17:23:14 <rajdeep> well we can generate on the horizon side as a work around ..
17:23:37 <thinrichs> I'll ask again: why can't we render rows without IDs?
17:23:45 <rajdeep> not sure about the horizon behaviour but i struggled with this
17:24:03 <rajdeep> janet's comment : # Policy table rows currently don't have ids. However, the DataTable object 	128     # requires an id for the table to get rendered properly. Otherwise, the 	129     # same contents are displayed for every row in the table. Assign the rows 	130     # ids here
17:24:43 <thinrichs> That sounds like a limitation of Horizon.  I'd agree with your suggestion rajdeep and gen them within horizon.
17:24:52 <pballand> thinrichs: +1
17:25:03 <rajdeep> ok thinrichs
17:25:49 <thinrichs> sarob: want to pick up again?
17:26:02 <pballand> longer term, I think we need to think more about the row strategy that congress exposes, but that shouldn’t block the UI work
17:26:44 <sarob> etherpad for the kilo design summit is here #link https://etherpad.openstack.org/p/par-kilo-congress-design-session
17:27:03 <sarob> im working out our space(s) at the summit
17:28:40 <sarob> im going to update the wiki to reference both new specs and minimum features
17:29:12 <glebo> Hey all, Gregory Lebovitz here. First time in the wkly mtg
17:29:13 <sarob> thats it other than y'all can start updating the wiki at will
17:29:38 <glebo> question, once we done talking about summit
17:29:39 <sarob> morning glebo
17:29:43 <thinrichs> glebo: Hi!  We'll ask you to give a brief intro near the end of the meeting.
17:29:54 <glebo> grt
17:30:56 <thinrichs> sarob: thanks!
17:31:08 <thinrichs> madhumohan: want to go next?
17:31:48 <madhumohan> been working on tempest tests for ceilometer driver....
17:32:59 <madhumohan> thinrichs: I am reviewing your code... half way now. will let you know if i find something interesting...
17:33:21 <thinrichs> madhumohan: great!  We'd love feedback on writing tempest tests.
17:33:33 <thinrichs> And having you review is really helpful.
17:34:13 <thinrichs> Anything else madhumohan?
17:34:43 <madhumohan> I had a policy where i needed a count of records in a table
17:35:50 <madhumohan> For eg: "A specific host can host utmost 6 VMs" OR "The subnet x can take utmost 10 ports"... any thoughts on supporting such constructs on datalog
17:35:52 <madhumohan> ?
17:36:14 <thinrichs> madhumohan: that would require aggregates, which we don't yet support.
17:36:33 <thinrichs> madhumohan: want to add a spec for it?
17:37:16 <madhumohan> yes.. I would like to take it up
17:38:27 <thinrichs> Sounds good.  I'll help.  That's a good suggestion.
17:38:52 <thinrichs> Starting to run short on time it seems.
17:38:54 <thinrichs> I'll go.
17:39:06 <thinrichs> I've been adding column-references to datalog,
17:39:27 <thinrichs> So we can write something like p(x) :- nova:servers(id=x, name=y)
17:39:44 <thinrichs> I've also been reactivating a simple form of enforcement that a customer asked for.
17:40:15 <thinrichs> So we can guard API calls using Congress.  e.g. before Nova executes a createVM API call, it can ask Congress if doing so would violate policy.
17:40:26 <thinrichs> The code is already in there but we deactivated it for the alpha release.
17:40:49 <thinrichs> I've also been asked to add some datetime functionality into the builtins that we support
17:40:57 <thinrichs> That's about it from me.
17:41:18 <thinrichs> Radu_: we're finishing up status reports.  Want to go?
17:41:51 <Radu_> I wanted to give the column referances a test today but was running into some other driver errors  with my driver, not sure if it has to do with the patch or something unrelated.
17:42:25 <Radu_> Since last week I have made the code and demo/audio publically visable
17:42:34 <thinrichs> Radu_: okay.  Let us know if we can help or if you find a bug.
17:42:42 <Radu_> https://www.youtube.com/watch?v=ZAEydTlIW64
17:43:07 <Radu_> Thats my demo video I made using my driver with congress to use congress as a reactionary tool for networks
17:43:56 <Radu_> Code repo is in the description. I also submitted a blueprint and a specs request to have the driver merged into the main repo
17:44:12 <thinrichs> It's great now that there's audio!
17:44:33 <thinrichs> Did you push your driver to review?
17:44:48 <Radu_> just the specs doc and blueprint
17:45:26 <thinrichs> If your driver is ready, go ahead and push it for review.
17:45:36 <thinrichs> We can go over both the specs and the driver at the same time.
17:46:21 <thinrichs> One other thing that I almost forgot.
17:46:37 <Radu_> Cool, I want to do a bit more testing and see If it is just an error that is occuring with the column referance patch or something else that was released in another patch I hadn't tested with that the column patch is based off of
17:47:00 <thinrichs> We talked to the Graffiti team this week.
17:47:03 <thinrichs> https://wiki.openstack.org/wiki/Graffiti
17:47:12 <thinrichs> Radu_: sounds good.
17:47:37 <thinrichs> Graffiti is project geared toward adding metadata to the objects in OS.
17:47:57 <Radu_> We also may move to using congress on an full openstack deployment, but that is still in the works and Im still on devstack for now.
17:48:15 <thinrichs> For example, they are making it possible to say that Apache is installed on some particular Glance image.
17:48:40 <pballand> that’s fantastic
17:48:46 <thinrichs> We were in complete agreement that the projects are complementary.
17:48:57 <thinrichs> Congress policies can be richer if we utilize Graffiti metadata.
17:49:30 <thinrichs> So we can say something like, "every VM that has Apache installed must have port 80 open"
17:49:46 <thinrichs> Is there anyone else that has a status report to give?
17:49:58 <thinrichs> We have a couple of new people this week and I want to ensure they have time to introduce themselves.
17:50:28 <thinrichs> #topic New members
17:50:49 <thinrichs> glebo: would you care to introduce yourself and say a little about why you're interested in Congress?
17:51:06 <glebo> Hey. It's Gregory from vArmour. Recall I was at the Silicon Valley 2-day session. This is just my first time in the wkly mtg.
17:51:16 <glebo> Repeat from the 2-day mtg,
17:51:37 <glebo> our interest is in being able to take security policies
17:51:41 <pballand> hi glebo, thanks for joining the meeting
17:51:44 <Radu_> Cool,that definately sounds like it could have a lot of use.
17:51:45 <glebo> regarding access control,
17:51:53 <glebo> reachability,
17:51:57 <glebo> microsegmentation
17:52:27 <glebo> etc from congress and be able to actually enforce those policies for each of the VMs
17:53:02 <thinrichs> glebo: sounds cool!
17:53:07 <glebo> I'd see us as a consumer of Congress,
17:53:38 <glebo> but willing and able to contribute to ensure that the structures and registries and interfaces are there for the security enforcement side of policy declaration
17:54:11 <glebo> Thx for the warm welcome. <hands off to others>
17:54:12 <jasonsb> glebo: that sounds really neat
17:54:13 <thinrichs> Makes sense.  We'd love to understand what your requirements are, so feel free to ping us anytime.
17:54:24 <thinrichs> jasonsb: want to take a turn?
17:54:26 <glebo> ack
17:54:52 <glebo> actually, I had a question…
17:55:00 <jasonsb> sure.  Im mostly lurking at the moment, trying to wrap my head around QoS
17:55:34 <thinrichs> You're certainly welcome to lurk anytime.
17:55:44 <jasonsb> and if congress can be used to express policies around quality of service as it relates to network and storage
17:55:50 <glebo> has there been any bp's or specs written around the interaction between congress and neutron GBP?
17:55:59 <thinrichs> But most of the weekly IRC are like what you're seeing—status updates.
17:56:04 <LouisF> thinrichs: may have been mentioned before but I am adding a spec and bp on translation from congress to gbp
17:56:16 <thinrichs> jasonsb: I'd suggest we communicate outside of this meeting to look into QoS policies.
17:56:24 <jasonsb> oh, and i've just started couple weeks back at HDS
17:56:42 <glebo> LouisF: well, there we have it then. Timely answer.
17:56:50 <glebo> ;-)
17:57:00 <jasonsb> thinrichs: sounds good.  i'm still early stages of wrapping my head around both congress and the problem domain
17:57:42 <thinrichs> jasonsb: let us know if you want pointers to intro material or want to chat on the phone.
17:58:00 <thinrichs> glebo: same for you, of course, though you've seen much of the intro material we have.
17:58:01 <jasonsb> LouisF: if you could point me to what you are working on i would like to see
17:58:31 <LouisF> jasonsb: i will post soon
17:58:37 <jasonsb> thinrichs: thank you for warm welcome
17:59:01 <jasonsb> i'll be in paris so would like to participate on your discussions as well
17:59:21 <thinrichs> sarob: can you post to the ML our Paris schedule?
17:59:28 <glebo> thinrichs: ack. thx.
17:59:37 <sarob> sure, when i get it locked down
17:59:42 <thinrichs> Sounds good.
17:59:57 <thinrichs> One last thing before we end.
18:00:15 <thinrichs> All the outstanding specs currently in the juno folder will be moved to the kilo folder.
18:00:26 <thinrichs> We can debate them all in Paris to prioritize.
18:00:31 <thinrichs> And with that we're out of time.
18:00:34 <thinrichs> Thanks all!
18:00:43 <LouisF> i am adding the gbp translation spec to the kilo folder
18:00:50 <thinrichs> #endmeeting