17:00:38 <thinrichs> #startmeeting CongressTeamMeeting
17:00:39 <openstack> Meeting started Tue Jul 22 17:00:38 2014 UTC and is due to finish in 60 minutes.  The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:42 <openstack> The meeting name has been set to 'congressteammeeting'
17:00:47 <thinrichs> Hi all
17:01:04 <banix> hi
17:01:15 <arosen> Hi
17:01:37 <skn_> Hi guys, sorry could not join last week's meeting
17:01:48 <thinrichs> skn_: no worries
17:02:07 <rajdeep> hi all
17:02:10 <thinrichs> pballand is out this week
17:02:18 <thinrichs> rajdeep: hi
17:02:23 <thinrichs> cloudtoad: you joining?
17:02:37 <thinrichs> sarob: Congress time
17:02:40 <rajdeep> hi thinrichs
17:02:43 <skn_> Did we move all the specs to congress-spec?
17:03:00 <thinrichs> skn_: let's wait for sarob to answer that
17:03:08 <skn_> ok
17:03:44 <thinrichs> I managed to put the agenda up on the google doc.
17:03:48 <thinrichs> https://docs.google.com/a/vmware.com/document/d/1p0-3mqtvtm7Yf3KnyrhuX_JMZJnrGlJhs2RoJkIezvI/edit#
17:04:08 <thinrichs> I guess we'll start with status updates
17:04:33 <skn_> why can't I access the doc?
17:04:41 <thinrichs> Hmmm… everyone with action items from last week hasn't arrived yet.
17:04:56 <arosen> skn_:  nope not yet but soon.
17:05:12 <skn_> ok
17:05:20 <thinrichs> skn_: not sure—let me see if I can change rights.
17:05:25 <arosen> skn_:  there was a bug in the config file for create the congress-specs repo so we didn't have gatting setup there. We're waiting till that patch merges before we move everything out.
17:05:34 <skn_> thinrichs: thanks
17:05:35 <rajdeep> thinrichs did you create it with your vmware id
17:05:56 <sarob> im here
17:06:04 <sarob> sorry double duty with the board meeting
17:06:09 <skn_> arosen: Got it
17:06:16 <thinrichs> sarob: glad you could make it
17:06:25 <sarob> i just got the congress-specs acls merged
17:06:27 <thinrichs> pballand made the doc.  I don't see how to make it publically avail.
17:06:41 <sarob> thinrichs::)
17:06:58 <sarob> congress-specs should be up on running today
17:07:06 <thinrichs> sarob: great!
17:07:08 <rajdeep> you can't
17:07:18 <sarob> i will start migrating the existing specs and idenifying gaps
17:07:25 <skn_> thinrichs: May be, you can just copy and upload a with your permission or something :)
17:07:55 <thinrichs> Let's let sarob report while he's able.  We can deal with agenda, etc. later.
17:08:20 <thinrichs> sarob: I see 2 other action items for you.  The policy-summit and the mid-cycle operators summit.  Anything to report?\
17:09:29 <thinrichs> Maybe sarob got pulled away.
17:09:58 <thinrichs> Agenda was simple: (i) status updates and (ii) a couple of discussion items: Tetris (another policy project) and some things I've learned while integrating)
17:10:08 <skn_> thinrichs: the agenda was for the policy summit?
17:10:17 <thinrichs> skn_: for this meeting
17:10:30 <thinrichs> I believe the policy-summit agenda is simple so far.
17:10:31 <sarob> sorry back
17:10:37 <thinrichs> Day 1: everyone talks about their policy projects
17:10:42 <skn_> oh shoot, I thought the policy summit
17:10:44 <thinrichs> Day 2: we discuss integrating
17:10:52 <sarob> policy summit is set for 18-19 sep
17:11:08 <skn_> thinrichs: Sounds right to me
17:11:15 <sarob> likely vmware location
17:11:27 <sarob> im going to coordinate with pierre
17:11:33 <sarob> from vmware
17:11:38 <sarob> location wise
17:11:42 <skn_> sarob: SFO?
17:11:48 <thinrichs> sarob: sounds good
17:11:50 <sarob> palo alto
17:11:56 <skn_> ok
17:12:21 <sarob> the operators summit, we likely will part of the working session part
17:12:41 <sarob> still working out with tom
17:12:42 <thinrichs> for the operators summit, do we give a talk with slides, a demo, a poster, what?
17:13:00 <sarob> its very crowded for topics
17:13:14 <skn_> operator summit, you mean the San Antonio one?
17:13:18 <sarob> so we are going to be part of the working session
17:13:25 <sarob> skn_ yup
17:14:03 <sarob> any questions?
17:14:11 <skn_> what's the working session supposed to contain?
17:15:16 <sarob> blueprints, telco, enterprise, ops tools
17:15:30 <sarob> collection of other stuff
17:15:58 <thinrichs> sarob: so we get a desk and can do a demo, put up a poster?  Is that the idea?
17:16:00 <skn_> so, we want to put in our blueprints in there? or something else?
17:16:30 <sarob> skn_: its more general on the ops interaction with blueprints i believe
17:16:54 <sarob> #link https://docs.google.com/spreadsheets/d/1p7gCZcLhmYnMC_Ab2N23Nndy3pAnTmtvpBvcvaOZb0Q/edit#gid=0
17:17:10 <sarob> this is the prelim ops mini-summit schedule
17:17:59 <skn_> which session could we be part of?
17:18:17 <thinrichs> Monday 1330 - 1415
17:18:18 <sarob> monday 25th aug 1330-1500
17:18:27 <sarob> yup
17:18:29 <thinrichs> called "Working Sessions"
17:18:35 <sarob> its breakout groups
17:19:06 <skn_> Oh okay
17:19:24 <sarob> it is a really busy couple of days
17:19:34 <skn_> So we get to present along with other WIP projects?
17:20:03 <sarob> this is not really for projects, rather ops topics
17:20:28 <sarob> like storage is general rather than ceph, swift, or cinder
17:20:54 <skn_> sarob: ok.  So what do you think our plan should be?
17:21:34 <sarob> overview of congress, strategy working with mulitple data sources, demo
17:22:21 <skn_> sarob: oh okay, got it.
17:22:51 <sarob> ill work to identify a few priority ops users if we can wrangle their attention for the working session
17:22:55 <sarob> show and tell time
17:23:01 <sarob> get some feedback
17:23:10 <thinrichs> that'd be great!
17:23:36 <skn_> sarob: sounds like a plan
17:23:49 <thinrichs> Anything else to discuss around either of the summits?
17:23:55 <skn_> I can help out on slides
17:24:08 <sarob> skn_: sounds great
17:24:41 <sarob> we can start dig details for the policy summit
17:24:53 <sarob> i wanted to get the specs going first
17:25:03 <sarob> so later this week, we start
17:25:39 <thinrichs> sarob: sounds good.  Let's make sure we do our best to get everyone working on policy an invite.
17:26:13 <skn_> Who are we inviting for the policy summit, other than we Congress guys? Did you reach out to them?
17:26:13 <sarob> okay
17:26:58 <sarob> i have reached out to the dev ML and directly to a few PTLs
17:26:59 <thinrichs> skn_: sarob is sending invites to a list of contacts.
17:27:23 <skn_> sarob: That's awesome! PTLs would be great
17:27:27 <sarob> once we have the agenda set, I will get more aggressive
17:28:03 <skn_> Let me know how I could help you, we don't want to burn you out
17:28:49 <skn_> Please CC/BCC us too ;)
17:29:51 <thinrichs> I think it's time to move on.
17:30:01 <thinrichs> Other status updates...
17:30:03 <skn_> thinrichs: yes, please.
17:30:08 <thinrichs> arosen: want to report what you're up to?
17:30:19 <arosen> So pretty much on the same path as last week.
17:30:35 <arosen> I have the keystone integration and policy.json integration up for review.
17:31:04 <arosen> I should pretty much be down with this soon. I want to add a file that describes how this integration is done. Once i do that hopefully we can merge it
17:31:26 <arosen> definitely eager to get some reviews on it but most of the code is boiler plate stuff that i stole from neutron/nova/keystone.
17:31:32 <arosen> that's it from me.
17:31:35 <kudva> Hi
17:31:39 <kudva> Sorry, I joined late
17:31:39 <thinrichs> It'd be great if we could spread out the reviewing some more.
17:31:44 <thinrichs> kudva: hi
17:31:48 <kudva> Hi
17:32:10 <thinrichs> I'm not especially well-versed in the OS way, so having banix and skn_ do some reviews of this keystone integration could be good.
17:32:29 <skn_> Sure, I could review
17:32:29 <gokul> hello
17:32:30 <banix> thinrichs: there were a lot of changes happening last week; now seem to be in a more stable state?
17:32:37 <banix> thinrichs: yes sure will do
17:32:46 <thinrichs> banix: yes—certainly more stable
17:33:16 <thinrichs> kudva: now that you're here, want to give us a status update?
17:33:26 <thinrichs> gokul: hello
17:33:43 <gokul> hi thinrichs:  hi -- apologize for the delay.  i forgot the meeting time.
17:33:54 <sarob> it would be great to get each contributor doing 2-3 reviews per day
17:33:55 <thinrichs> gokul: no worries—glad you made it!
17:33:57 <kudva> Last week was crazy for me since I had to give a talk elsewhere. Not much update this week. next week, I will have some
17:34:08 <thinrichs> sarob: agreed
17:34:26 <thinrichs> kudva: is the builtin code ready for review?
17:34:46 <kudva> The core builtin code and the integration both are
17:35:21 <kudva> The test environment changed (by arosen), before I was using the tests in scripts, would like to do the new tests before submitting. In a couple of days
17:35:37 <rajdeep> i am having issues getting alert emails from gerrit
17:35:51 <rajdeep> how do we track CLs coming in for review
17:35:58 <arosen> rajdeep:  if you go under settings in gerrit you can change your alert settings.
17:36:21 <thinrichs> kudva: OK—let me know when they're ready.  I'm guessing there will be a couple of other things to do: (i) add syntax checkers to make sure people use those builtins properly, and (ii) change our query "optimizer" to order those builtins properly.
17:37:11 <thinrichs> rajdeep: I haven't found a better way to track what reviews are outstanding other than looking at the list on gerrit.
17:37:15 <kudva> thinrichs: (i) is clear, what do you mean by (ii)
17:37:46 <thinrichs> kudva: maybe discuss this offline…it's related to (i)
17:37:55 <kudva> thinrichs: okay
17:38:30 <thinrichs> 2 other topics to discuss that might eat up time.  Anyone else have status updates?
17:38:52 <skn_> Yes, I have started working on the IDS use case for Congress
17:38:59 <rajdeep> i submitted  a CL to fix neutron driver test case which was failing
17:39:22 <skn_> I am going to own this up for the Juno summit
17:39:46 <thinrichs> rajdeep: I have a bunch of changes to the neutron driver during my integration work.  We should sync your change with mine.
17:40:01 <rajdeep> ok..
17:40:03 <thinrichs> skn_: can you give us more details?
17:40:08 <skn_> The first thing I am working on is to add support to Neutron for tap mode
17:40:49 <skn_> "tap" mode gives us a way to forward L2 traffic to an IDS VM
17:41:11 <banix> skn_: do you a spec under review in Neutron?
17:41:14 <thinrichs> How is that related to the work on service-chaining in neutron?
17:41:55 <skn_> I am creating a spec, will have to put it in Neutron
17:42:04 <banix> there is ongoing discussions under the Advanced Services that also include a proposal for TAPaaS
17:42:31 <banix> skn_: need to make sure you are aware of the related work happening in Neutron; we an talk offline
17:42:40 <skn_> banix: Who can I talk to about that?
17:42:54 <rajdeep> TAPaas sounds scary
17:43:03 <skn_> Yes
17:43:03 <rajdeep> :)
17:43:07 <banix> skn_: start from me :)
17:43:28 <skn_> banix: hahaha sure.  Let us talk offline on this then
17:43:32 <banix> There is an active working group looking into these topics
17:44:12 <banix> skn_: just to avoid replicating work, etc….
17:44:22 <skn_> banix: sure
17:44:32 <thinrichs> OK.  Time to move on.  We have someone new joining us.
17:44:53 <banix> drum rolls
17:44:56 <thinrichs> We had a discussion with the Tetris group last week, of which gokul is a member.
17:45:11 <thinrichs> gokul: want to say a few words about Tetris and why you're here?
17:46:34 <thinrichs> Looks like we lost gokul.  Hopefully he'll be back.
17:46:53 <banix> yeah looks like network disconnection
17:47:01 <thinrichs> I'll see if I can give a brief overview, and he'll correct me when he gets back.
17:47:35 <thinrichs> Tetris is a policy project aimed at enforcing policy that includes optimization problems.
17:47:55 <thinrichs> It has the same kind of scope as Congress: working with any collection of datacenter services.
17:48:15 <thinrichs> So when we talked on the phone, it quickly became clear that Congress and Tetris had very similar goals.
17:48:42 <skn_> Who are all working on Tetris?
17:48:42 <thinrichs> In the end, we decided to combine efforts.
17:49:06 <thinrichs> gokul is a better person to ask than me.  kudva: do you know?
17:49:29 <banix> gokul and Kudva have lost connectivity
17:49:46 <thinrichs> Here's their meeting web page.
17:49:49 <thinrichs> #link https://wiki.openstack.org/wiki/Meetings/Tetris
17:50:16 <thinrichs> You can see attendees at the top.
17:51:05 <skn_> thinrichs: thanks
17:51:11 <thinrichs> In short, having them on board will be really helpful when we move from focusing on monitoring to focusing on enforcement.
17:51:20 <skn_> kudva is in that meeting, it looks like
17:52:17 <thinrichs> Speaking of monitoring and enforcement, I'm starting to believe than monitoring is going to require more work than we had thought.
17:53:01 <thinrichs> Here's the lowdown on what I learned from my integration work.
17:53:23 <thinrichs> 1. Having tables with lots of columns to represent say Neutron data makes it hard to write policies.
17:53:51 <thinrichs> Here's one of the tables we currently have.
17:53:53 <thinrichs> neutron:ports(addr_pairs, security_groups, extra_dhcp_opts, binding_cap, status, name, admin_state_up, network_id, tenant_id, binding_vif, device_owner, mac_address, fixed_ips, port, device_id, binding_host_id1)
17:54:12 <thinrichs> This is the ports table—it's got 16 columns.
17:54:56 <thinrichs> I made 3 separate mistakes writing a simple rule because I had variables in the wrong order, used the same variables in two different predicates (an accidental 'join'), and had the wrong number of arguments.
17:55:23 <thinrichs> So in short, I'm trying to figure out how we can make these data sources easier to use for policy writers.
17:55:28 <skn_> Why do we have to specify all the fields if we are not using them in a policy, for instance?
17:55:44 <thinrichs> If people can't write policy easily, they won't.
17:55:51 <sarob> because they are not objects?
17:56:13 <sarob> rather than list
17:56:17 <thinrichs> The abstraction we're imposing on datasources is that they export tables.
17:56:27 <sarob> array
17:56:27 <thinrichs> where "table" means "database table".
17:56:52 <skn_> thinrichs: I see
17:56:54 <thinrichs> Perhaps that's the wrong abstraction.  The reason we chose it is to ensure we can leverage database technology to do policy evaluation.
17:57:08 <thinrichs> DB technology is good b/c we know we can scale it pretty well.
17:57:30 <thinrichs> The question is how we make it easier to write policy yet still retain DB tech part.
17:57:57 <skn_> thinrichs: we'll need a layer in between to verify/infer facts
17:58:02 <thinrichs> I have some ideas.  Anyone want to work through this?
17:58:12 <sarob> thinrichs: yes
17:58:27 <sarob> 2 minutes
17:58:33 <skn_> we are running out of time
17:58:37 <thinrichs> I'll put together some options with pros/cons and circulate it.
17:58:37 <skn_> yes
17:58:41 <sarob> can move to congress channel
17:58:52 <skn_> thinrichs: that'll be great
17:58:53 <sarob> thinrichs: that would be great
17:59:18 <thinrichs> OK.  Anyone wanting to continue, we will jump to #congress.
17:59:34 <thinrichs> Otherwise, we'll see you next week!
17:59:43 <banix> bye
18:00:04 <thinrichs> #endmeeting