18:01:55 <SumitNaiksatam> #startmeeting networking_policy
18:01:56 <openstack> Meeting started Thu Jul 16 18:01:55 2015 UTC and is due to finish in 60 minutes.  The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:01:57 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
18:02:00 <openstack> The meeting name has been set to 'networking_policy'
18:02:07 <songole> hi
18:02:18 <SumitNaiksatam> #info agenda https://wiki.openstack.org/wiki/Meetings/GroupBasedPolicy#July_16th.2C_9th_2015
18:02:27 <Yi> hi
18:02:31 <SumitNaiksatam> try to get to some of the items we did not cover in last weeks agenda
18:02:33 <SumitNaiksatam> Yi: hi
18:02:55 <SumitNaiksatam> update on the kilo release - we are still working on it
18:04:14 <SumitNaiksatam> we have users who are also testing this, so it would be good to incorporate their input and provide any major fixes before we release
18:04:37 <SumitNaiksatam> #topic Bugs
18:05:05 <SumitNaiksatam> its not a bug, but a criticial issue is that the GBP py27 gate is broken
18:05:30 <SumitNaiksatam> this happened after there was an update to the global requirements yesterday
18:05:42 <ransari> @SumitNaiksatam: which bug are you referring to?
18:05:49 <SumitNaiksatam> i have patch to update our test-requirements (thanks mageshgv for your input):
18:06:03 <SumitNaiksatam> #link https://review.openstack.org/202668
18:06:11 <SumitNaiksatam> ransari: did not file it as a bug
18:06:24 <SumitNaiksatam> since update of requirements is an ongoing process
18:06:47 <ransari> @SumitNaiksatam ok
18:06:54 <SumitNaiksatam> the patch has passed the py27 gate now, but waiting on the integration test
18:07:05 <SumitNaiksatam> the upstream gate is pretty busy and backed up
18:07:06 <igordcard> SumitNaiksatam: I was just testing that patch now
18:07:10 <SumitNaiksatam> igordcard: ah ok
18:07:13 <SumitNaiksatam> igordcard: thanks
18:07:27 <SumitNaiksatam> one thing to note is that with the new hacking requirements, pep8 was failing on our code
18:07:45 <SumitNaiksatam> so i slyly reverted back to the earlier hacking version
18:08:09 <SumitNaiksatam> in a separate patch we should update the hacking version, and fix those pep8 issues
18:08:45 <SumitNaiksatam> requesting the team to keep a watch on the above patch and help merge when you see it pass the integration job
18:09:02 <SumitNaiksatam> we cant merge other patches until that fix merges
18:09:18 <SumitNaiksatam> i dont think we are tracking any other critical priority bugs this week
18:09:27 <SumitNaiksatam> any other bugs anyone wants to bring up?
18:10:30 <mageshgv> #link https://review.openstack.org/#/c/198052/
18:10:31 <SumitNaiksatam> okay so moving on
18:10:49 <SumitNaiksatam> oh
18:11:13 <mageshgv> We have Service Node configuration as updateable parameter in the spec, but in implementation it is different today
18:11:21 <SumitNaiksatam> mageshgv: this is #link https://launchpad.net/bugs/1470866
18:11:21 <openstack> Launchpad bug 1470866 in Group Based Policy "Service node config should be an updateable field" [High,Confirmed] - Assigned to Magesh GV (magesh-gv)
18:11:35 <ransari> @SumitNaiksatam important because without this support, service node config requires disruptive re-deploy of the chain
18:11:48 <SumitNaiksatam> mageshgv: ransari: okay
18:12:03 <SumitNaiksatam> mageshgv: i provided some minor comments
18:12:25 <SumitNaiksatam> other than that, i believe the ratinale is that if a driver cannot support an update it will throw an exception?
18:12:31 <SumitNaiksatam> *rationale
18:12:34 <ransari> @mageshv: you mentioned it needs to be updated ?
18:12:42 <mageshgv> SumitNaiksatam: Thanks. Will post an updated patch once our py27 gets fixed
18:12:51 <SumitNaiksatam> mageshgv: ok
18:13:10 <mageshgv> SumitNaiksatam: yes, a driver that does not support update will reject it
18:14:07 <SumitNaiksatam> just a general comment that in the specs we are using different conventions to describe our resrouces
18:14:21 <SumitNaiksatam> in #link https://github.com/stackforge/group-based-policy-specs/blob/master/specs/juno/group-based-policy-abstraction.rst
18:14:34 <SumitNaiksatam> we have the resource map
18:14:54 <SumitNaiksatam> but in #link https://github.com/stackforge/group-based-policy-specs/blob/master/specs/juno/group-based-policy-service-chaining.rst
18:15:02 <SumitNaiksatam> we use a tabular convention
18:15:09 <SumitNaiksatam> confusing for the reader and also to track
18:15:24 <SumitNaiksatam> in the later we use “W” to indicate both create and update
18:16:09 <SumitNaiksatam> anyone have any objections with allowing support of config updates to instantiated nodes?
18:16:27 <SumitNaiksatam> i am referring to #link https://review.openstack.org/#/c/198052/
18:16:54 <SumitNaiksatam> ok
18:17:01 <SumitNaiksatam> any other bugs to discuss?
18:17:30 <SumitNaiksatam> moving on
18:17:34 <SumitNaiksatam> #topic Integration tests
18:17:39 <SumitNaiksatam> just a quick update here:
18:18:09 <SumitNaiksatam> in #link https://review.openstack.org/#/c/202263/ i am experimenting with adding rally tests
18:18:21 <SumitNaiksatam> ideally this should not be a part of the same integration job
18:18:37 <SumitNaiksatam> but we are already setting up devstack, so i will try and see how this behaves
18:19:13 <SumitNaiksatam> the rally tests are being pulled from here: #link https://github.com/group-policy/rally/tree/dev
18:19:23 <SumitNaiksatam> written by ajay (we had discussed this a few weeks back)
18:19:50 <SumitNaiksatam> comments/questions?
18:20:17 <SumitNaiksatam> #topic Packaging Update
18:20:31 <SumitNaiksatam> rkukura: anything you want to update the team on?
18:20:47 <rkukura> nothing new this week
18:20:54 <SumitNaiksatam> perhaps based on your earlier meeting today ;-)
18:21:29 <rkukura> I did meet with some Red Hat people today, but we didn’t really cover much specific to GBP
18:21:59 <rkukura> I think we can get momentum going again on that, though
18:22:14 <SumitNaiksatam> rkukura: okay
18:22:24 <SumitNaiksatam> #topic Docs
18:22:37 <ransari> @rkukura: do we expect updated Juno rpms?
18:22:46 <SumitNaiksatam> #undo
18:22:47 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x87652d0>
18:23:38 <rkukura> I still hope to do those, but have been spending way too much time trying to get the nova node driver to work right
18:24:28 <SumitNaiksatam> rkukura: you can let the team know if we can help in any way on the rpms
18:24:50 <ransari> @rkukura: thnaks
18:25:04 <SumitNaiksatam> #topic Docs
18:25:12 <SumitNaiksatam> so we have #link https://github.com/stackforge/group-based-policy/tree/master/doc/source
18:25:28 <SumitNaiksatam> the way most other projects are using this is for “devref”
18:25:40 <SumitNaiksatam> e.g.: in nova #link https://github.com/openstack/nova/commit/47b233f988bddd1272e4a1bc02be97f15701a8ca
18:26:01 <SumitNaiksatam> we currenly have GBP overview here
18:26:23 <SumitNaiksatam> it will help to have devref for things like writing node drivers for the new node composition plugin
18:26:34 <SumitNaiksatam> just wanted to bring it to everyone’s attention
18:26:50 <SumitNaiksatam> currently we are lacking sorely on good API documentation
18:27:02 <SumitNaiksatam> and on good developer documentation
18:27:18 <SumitNaiksatam> so if you have ideas on how do this, and/or want to contribute, please let me know
18:27:37 <SumitNaiksatam> on a related note, we are also missing a python SDK for GBP
18:28:24 <ransari> @umitNaiksatam I can help out
18:28:25 <SumitNaiksatam> similar to #link https://wiki.openstack.org/wiki/SDK-Development/PythonOpenStackSDK
18:28:33 <SumitNaiksatam> ransari: that will be great
18:28:43 <rkukura> SumitNaiksatam: What is needed for a python SDK beyond python-gbpclient?
18:29:13 <rkukura> mainly docs?
18:29:46 <SumitNaiksatam> rkukura: i agree one case use the client lib, but people like to have a more evolved library to work with
18:29:55 <SumitNaiksatam> this was the feedback from a few uses
18:29:57 <SumitNaiksatam> *users
18:30:05 <SumitNaiksatam> and, yes, it is self documenting
18:30:09 <rkukura> so a layer above python-gbpclient?
18:30:13 <SumitNaiksatam> rkukura: yeah
18:31:14 <SumitNaiksatam> rkukura: python-openstacksdk has python bindings for most openstack projects (which were earlier part of the integrated release)
18:31:31 <rkukura> its not clear to me if this uses the python-*client libs, or is something different. Also, how is it related to the old OpenStack client effort?
18:31:51 <SumitNaiksatam> rkukura: this would be different from the openstack client efforts
18:32:20 <SumitNaiksatam> and my understanding is that it would use the client libs
18:33:13 <SumitNaiksatam> #link https://github.com/stackforge/python-openstacksdk/tree/master/openstack
18:33:16 <SumitNaiksatam> ok moving on
18:33:30 <SumitNaiksatam> #topic Stable branches for clients
18:33:48 <SumitNaiksatam> i checked with the folks in the team if they were okay with this idea
18:33:58 <SumitNaiksatam> other projects are already doing it
18:34:15 <SumitNaiksatam> most everyone seems to be in agreement
18:34:30 <SumitNaiksatam> this will allow us to backport fixes
18:34:59 <SumitNaiksatam> if there is no disagreement, i can go ahead and create stable client branches for juno and kilo
18:35:16 <SumitNaiksatam> this does not affect the client version numbering
18:35:56 <SumitNaiksatam> rkukura: should we use 1.0.x for stable/kilo?
18:36:37 <rkukura> SumitNaiksatam: I have not looked into how the other client library stable branches are being versioned, so I’m not sure
18:37:37 <SumitNaiksatam> rkukura: i am mostly asking (and this is to the team in general), if we can transition to client version 1.0 (currently we are 0.9.x)
18:37:59 <SumitNaiksatam> we dont have to decide now, but we will soon get to that point
18:38:17 <rkukura> I’d think so - we can always go to 2.x if/when we make incompatible changes
18:38:26 <SumitNaiksatam> rkukura: yeah
18:38:34 <ransari> @SumitNaiksatam: good to have back ports to juno client branch
18:39:09 <rkukura> Are there API changes in kilo/master that we have not and are not backporting to juno?
18:39:11 <SumitNaiksatam> ransari: okay, we will update the minor version number to reflect that
18:39:43 <SumitNaiksatam> rkukura: there at least a couple i think of
18:39:49 <ransari> @rkukura: We haven't back ported service profile changes to juno based client
18:40:02 <SumitNaiksatam> ransari: yes, that is one of those tow
18:40:07 <SumitNaiksatam> *two
18:40:08 <rkukura> OK
18:40:24 <SumitNaiksatam> the other one is with mageshgv’s patch we discussed earlier today
18:40:35 <SumitNaiksatam> there is at least a third one -
18:41:08 <ransari> @SumitNaiksatam: mageshv patch is not merged in Kilo
18:41:18 <SumitNaiksatam> ransari: yes, but we expect it will
18:41:23 <SumitNaiksatam> when creating or updating a group there is a provide and consume attribute
18:41:25 <SumitNaiksatam> this is a dict
18:41:37 <SumitNaiksatam> but the format of the dict is confusing to most users
18:41:39 <ransari> @SumitNaiksatam: ok .
18:41:43 <SumitNaiksatam> (the use of scope)
18:42:23 <SumitNaiksatam> so we are planning to change that, though the change can be done in a backward compatible way (such that you can support the old format as well for one cycle of deprecation)
18:42:44 <SumitNaiksatam> a fourth change would have been to rename external-policy to external-group
18:43:24 <ransari> yes, use of scope is confusing
18:44:00 <SumitNaiksatam> in general, rkukura per your question, this might be a good time to indentify (and perhaps the last chance) to introduce a changes to the API
18:44:31 <rkukura> right
18:45:07 <SumitNaiksatam> so everyone please thing through, and if you have any issues with the current usage, please dont hesitate to raise them
18:45:43 <SumitNaiksatam> #topic Node drivers
18:45:59 <SumitNaiksatam> igordcard: any updates you want to share?
18:46:51 <SumitNaiksatam> rkukura: anything at your end with the nova driver?
18:47:14 <rkukura> I’m kind of stuck with some weird lockup between neutron and nova
18:47:21 <igordcard> SumitNaiksatam: I am now more focused on TS than before, so you can expect news during this month
18:47:43 <SumitNaiksatam> igordcard: thats great
18:47:51 <SumitNaiksatam> rkukura: okay
18:48:20 <rkukura> Neutron stops accepting new requests, including those being made from nova.
18:48:34 <igordcard> the nova node driver would be helpful, but I will temporarily create a dummy node driver that receives port IDs already created, to automate debugging and development
18:48:44 <rkukura> Having neutron call nova in a separate thread does not seem to help
18:48:47 <SumitNaiksatam> igordcard: okay
18:49:05 <SumitNaiksatam> rkukura: ah, so the multiple api threads did not work, darn
18:49:19 <rkukura> that didn’t solve it either
18:49:48 <SumitNaiksatam> rkukura: so in your opinion this is because the nova driver is running inside the neutron process space?
18:50:17 <rkukura> the actual POST call to nova seems to return the instance ID to neutron, but then novaclient tries to do a get on that ID, and nova within that does a get_port, which neutron doesn’t respond to.
18:50:35 <rkukura> SumitNaiksatam: I have no theory on why it shoudn’t work inside neutron
18:50:45 <rkukura> I’m probably doing something blatently wrong
18:51:18 <SumitNaiksatam> rkukura: would it possible to post a WIP patch so folks in the team can look and perhaps provide some ideas?
18:51:49 <rkukura> Maybe - the little bit of code is a real mess - I’ve really just been experimenting with different approaches
18:52:29 <SumitNaiksatam> rkukura: okay
18:52:34 <SumitNaiksatam> mageshgv: there are a few references to “reference” as regards the older “simple chain driver”, we should remove those
18:52:38 <rkukura> I’ve got a couple more ideas to try, but will try to post something for others to look at if I don’t get it solved this week
18:52:46 <SumitNaiksatam> rkukura: thanks
18:53:21 <mageshgv> SumitNaiksatam: ok
18:53:50 <SumitNaiksatam> i think Ivar had to leave for another meeting, so we will skip his patches (but please do review them)
18:53:56 <SumitNaiksatam> #topic Open Discussion
18:54:17 <SumitNaiksatam> the deadline for talk submissions to Tokyo was yesterday
18:54:35 <SumitNaiksatam> Ivar has submitted a proposal for the hands-on lab
18:55:10 <SumitNaiksatam> anything else anyone wants to discuss today?
18:55:28 <SumitNaiksatam> or we can wrap up early today ;-P
18:55:52 <SumitNaiksatam> alright, thanks everyone for attending!
18:56:02 <SumitNaiksatam> bye!
18:56:07 <rkukura> thanks SumitNaiksatam!
18:56:08 <mageshgv> bye
18:56:11 <SumitNaiksatam> #endmeeting