18:01:16 <SumitNaiksatam> #startmeeting networking_policy
18:01:17 <openstack> Meeting started Thu Sep  1 18:01:16 2016 UTC and is due to finish in 60 minutes.  The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:01:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
18:01:21 <openstack> The meeting name has been set to 'networking_policy'
18:01:29 <songole> hi SumitNaiksatam
18:01:35 <SumitNaiksatam> #info agenda https://wiki.openstack.org/wiki/Meetings/GroupBasedPolicy#Sept_1st_2016
18:01:38 <SumitNaiksatam> rkukura: hi
18:01:42 <rkukura> hi
18:01:51 <SumitNaiksatam> #topic Bugs
18:02:31 <SumitNaiksatam> quite a few bugs have been reported over the last week
18:02:46 <SumitNaiksatam> i am guessing these have surfaced from the NFP related testing
18:02:51 <SumitNaiksatam> #link https://bugs.launchpad.net/group-based-policy/+bug/1617873
18:02:51 <openstack> Launchpad bug 1617873 in Group Based Policy "Liberty Devstack installation failure" [Undecided,New]
18:03:03 <SumitNaiksatam> songole: do you know if vikash was able to make progress?
18:03:47 <tbachman> FYI — I saw this before
18:03:55 <SumitNaiksatam> tbachman: aha
18:04:01 <tbachman> I’m trying to remember what fixed it
18:04:12 <tbachman> It may have been just pulling a newer version of devstack/repos
18:04:15 <SumitNaiksatam> tbachman: to me it seemed like a devstack issue not GBP specific
18:04:20 <SumitNaiksatam> tbachman: right
18:04:20 <tbachman> SumitNaiksatam: ack
18:04:33 <SumitNaiksatam> tbachman: okay cool
18:04:37 <SumitNaiksatam> #link #link https://bugs.launchpad.net/group-based-policy/+bug/1619129
18:04:37 <openstack> Launchpad bug 1619129 in Group Based Policy "PRS is not detached when PTG update with that PRS failed" [Undecided,New]
18:04:42 <songole> i will check with him. Is this on Juno?
18:04:43 <songole> sorry, bug says liberty
18:05:33 <SumitNaiksatam> songole: on the second one: 1619129
18:05:51 <SumitNaiksatam> seems like a basic thing, i am surprised that this was not caught before
18:06:21 <SumitNaiksatam> and i am concerned that its a regression
18:06:46 <SumitNaiksatam> songole: is it the observation that PRS update does not happen at all for a PTG?
18:07:27 <SumitNaiksatam> songole: dont mean to put you on the spot, I can post the question on the reported bug in case you are not aware
18:07:34 <songole> It was just reported. I will find out.
18:07:59 <SumitNaiksatam> songole: okay
18:08:20 <songole> on the liberty issue with devstack
18:08:22 <SumitNaiksatam> #link https://bugs.launchpad.net/group-based-policy/+bug/1615033
18:08:22 <openstack> Launchpad bug 1615033 in Group Based Policy "Router interface corresponding to subnet of a PTG is not added to L3P router in explicit flow" [Undecided,New]
18:08:38 <SumitNaiksatam> songole: go ahead
18:08:43 <songole> vikash tried the 2nd time, and it worked. So, he was able to make progress
18:09:11 <songole> we could probably close it
18:10:17 <SumitNaiksatam> songole: ok good, if he can respond/update/close the bug it will be good
18:10:42 <SumitNaiksatam> regarding, 1615033 (^^^) do you know which policy drivers are being used
18:10:43 <SumitNaiksatam> ?
18:11:55 <SumitNaiksatam> tbachman: is this still an issue https://bugs.launchpad.net/group-based-policy/+bug/1605614 ?
18:11:55 <openstack> Launchpad bug 1605614 in Group Based Policy "Nova client v2 can't be used directly." [Undecided,New]
18:12:24 * tbachman goes to look at bug..
18:13:14 * igordcard was on the walk, says hi
18:13:34 <tbachman> SumitNaiksatam: I think this was fixed by another commit
18:13:44 <tbachman> maybe it was rkukura’s backport of his fix to tox.ini?
18:13:48 <tbachman> (to add requirements)
18:13:50 * SumitNaiksatam hopes igordcard enjoyed his walk ;-)
18:13:55 * tbachman wish he had a better memory
18:14:16 <SumitNaiksatam> tbachman: aha, its the one from 7/22, i confused it as being created in aug
18:14:24 <SumitNaiksatam> tbachman: yeah its the same one, nevermind
18:14:53 <SumitNaiksatam> the rest of the bugs seem to be enhancements to made for NFP (as followup to review comments)
18:15:04 <SumitNaiksatam> songole: hopefully those enhancements will happen at some time :-)
18:15:25 <SumitNaiksatam> any other critical bugs or pending patches that anyone would like to bring up before we move on?
18:15:47 <songole> SumitNaiksatam: yeah, most are done. need to update and close them
18:16:11 <SumitNaiksatam> songole: sweet!
18:16:59 <SumitNaiksatam> songole: it will be helpful if the bugs being fixed are referred to in the commit messages, that way they will be closed automatically when the merge happens
18:17:10 <SumitNaiksatam> songole: same patch can address multiple bugs
18:17:11 <songole> ok
18:17:14 <SumitNaiksatam> ok mving one
18:17:17 <SumitNaiksatam> songole: thanks
18:17:26 <SumitNaiksatam> #topic Packaging
18:17:29 <SumitNaiksatam> rkukura: hi
18:17:55 <rkukura> hi - nothing new from me on this
18:18:05 * tbachman wonders if rkukura is going to change his NIC to “packaging"
18:18:42 <SumitNaiksatam> anything to discuss with regards to the new RH proposed schedule that came up in the discussion yesterday?
18:18:46 <SumitNaiksatam> tbachman: lol!
18:19:03 <rkukura> SumitNaiksatam: I don’t think we are ready to talk about that
18:19:06 <SumitNaiksatam> TPG - the packaging guru!
18:19:33 <SumitNaiksatam> rkukura: okay, i was wondering if we needed to get anything ready from that perspective
18:20:20 <rkukura> SumitNaiksatam: Did you have an email conversation about packages?
18:20:22 <SumitNaiksatam> oh one note on that front - the packaging repo is now public and open for collaboration: #link https://github.com/group-policy/gbp-packages
18:20:29 <SumitNaiksatam> rkukura: no i dont
18:21:12 <SumitNaiksatam> so if you want to add any new packaging specs, support new distros, or make changes to existing to specs, please send a pull request to the above repo
18:21:25 <SumitNaiksatam> ping me if you face any issues
18:21:45 <SumitNaiksatam> #topic Testing
18:22:04 <SumitNaiksatam> our stable/juno branch was breaking in the upstream gate
18:22:32 <SumitNaiksatam> thanks to ivar-laz_ who figured out that this was due to the tests now running on xenial
18:23:24 <SumitNaiksatam> apicapi was pulling a ssl package on xenial that was not matchich what other openstack juno libs were expecting
18:24:28 <SumitNaiksatam> i am bringing this up so that if you see a similar wierd issue in the future, be sure to run your tests on xenial in your local env (and you might see that it behaves differently than your trusted trusty!)
18:25:06 <SumitNaiksatam> #topic Tags for GBP resources
18:25:14 <SumitNaiksatam> there is no spec for this yet
18:25:51 <SumitNaiksatam> but the basic idea is to do what is being already done in Neutron for the network resource:
18:25:53 <SumitNaiksatam> #link http://docs.openstack.org/mitaka/networking-guide/ops-resource-tags.html
18:26:30 <SumitNaiksatam> the tags will not be interpreted by the GBP API or plugin layer, and will be relayed to the driver
18:27:01 <SumitNaiksatam> the policy driver can make finer granularity policy decisions based on the tags
18:27:16 <SumitNaiksatam> the initial plan is to introduce this only for the PT resource
18:28:09 <songole> what is the use case for tags?
18:28:13 <SumitNaiksatam> i will post a spec, but i wanted to get the thought cycles going on this
18:28:36 <SumitNaiksatam> songole: microsegmentation has been suggested as one use case
18:28:51 <SumitNaiksatam> songole: and is applicable in the context of applying tags on PTs
18:29:43 <songole> neutron network doesn't map to PT,right?
18:29:59 <songole> or the other way..
18:30:00 <SumitNaiksatam> songole: no, PT maps to a port
18:30:32 <ivar-laz_> SumitNaiksatam: are these tags or labels?
18:30:35 <SumitNaiksatam> songole: hence its microsegmentation, not segmentation :-)
18:30:55 <SumitNaiksatam> ivar-laz_: in my mind, tags, labels, same difference
18:30:56 <songole> SumitNaiksatam: Is neutron providing it on ports as well?
18:30:58 <ivar-laz_> Who is going to understand them? Top layer orchestration or southbound drivers?
18:31:01 <SumitNaiksatam> songole: no
18:31:34 <ivar-laz_> SumitNaiksatam: in my mind Tags are usually interpreted by top level orchestration, there's no enforcement of them in the datapath
18:31:37 <SumitNaiksatam> ivar-laz_: that is up for discussion, but in the initial “exploratory” cut, the semantics are applied by the drivers
18:32:20 <ivar-laz_> SumitNaiksatam: I see... That is an API backdoor though
18:32:21 <SumitNaiksatam> ivar-laz_: in this case, the definition of tags is similar to the way Neutron defines it, in that Neutron really does not interpret these
18:33:01 <ivar-laz_> SumitNaiksatam: If Neutron doesn't interpret the tags, then the drivers shouldn't as well, right?
18:33:03 <SumitNaiksatam> ivar-laz_: i think they had similar discussions and concerns in neutron as well
18:33:12 <ivar-laz_> I haven't read the spec though, I'll give a look
18:33:43 <ivar-laz_> If you think of APIC tags for example, they are purely for northbound users to understand... They make no impact on the datapath
18:34:17 <ivar-laz_> But if the drivers are going to interpret them, they are basically a free-of-charge API extension layer
18:34:21 <SumitNaiksatam> ivar-laz_: as course of evolution these tags would be defined in GBP itself and would have semantics
18:34:30 <SumitNaiksatam> ivar-laz_: i wouldnt go that far
18:35:28 <ivar-laz_> A generic key-value storage that is interpreted by the drivers for changing datapath semantics, sounds like putting extra info in the description fields
18:35:52 <ivar-laz_> But I'll look at the Neutron spec, maybe I'm just imagining the wrong thing
18:36:31 <SumitNaiksatam> ivar-laz_: like i said, as patch of evolution we be defining the tags in GBP as well, the initial goal is to extend support for the Neutron feature to GBP
18:36:53 <SumitNaiksatam> patch -> path
18:37:43 <SumitNaiksatam> any other thoughts?
18:38:36 <SumitNaiksatam> #topic Application Policy Group
18:38:58 <SumitNaiksatam> #link https://review.openstack.org/#/c/339272
18:39:18 <SumitNaiksatam> posted this a while back but i am guessing not too many people had a chance to review this
18:39:40 <SumitNaiksatam> the idea is pretty simple, a grouping construct on top of PTGs
18:40:00 <rkukura> does this effect the behaviour of anything?
18:40:08 <SumitNaiksatam> but this enables adding policies to applications in the future
18:40:43 <rkukura> if not, why could the tags be used for the same purpose?
18:40:47 <SumitNaiksatam> rkukura: it doesnt in the current scheme of things, only something that is aware of the APG can make use of it
18:41:06 <rkukura> s/could/couldn’t/
18:41:58 <SumitNaiksatam> rkukura: could be, but this provides a more concrete construct by way of resource model and API
18:42:28 <rkukura> I’ll read the spec to see why this is required
18:42:35 <SumitNaiksatam> rkukura: okay, thanks
18:43:29 <SumitNaiksatam> any other thoughts/comments on this?
18:44:38 <SumitNaiksatam> #topic Open Discussion
18:44:47 <SumitNaiksatam> anything else anyone wants to bring up today?
18:45:20 <songole> SumitNaiksatam: NFP patches are posted for review
18:45:37 <SumitNaiksatam> songole: right
18:46:11 <SumitNaiksatam> #link https://review.openstack.org/#/q/topic:bp/gbp-network-services-framework
18:46:21 <SumitNaiksatam> the open patches
18:46:31 <SumitNaiksatam> not sure if anyone else had a chance to look at these
18:47:28 <SumitNaiksatam> songole: but if you and hemanthravi think they are fine, i am okay with proceeding with these since they are almost entirely contained to the NFP subsystem of the project
18:47:37 <SumitNaiksatam> anyone else have any objections?
18:47:48 <songole> SumitNaiksatam: ok
18:47:50 <SumitNaiksatam> please chime in here or on the reviews
18:47:57 <rkukura> no objection, but will at least try to skim through them over the next few days
18:48:13 <songole> rkukura: thanks
18:49:13 <SumitNaiksatam> rkukura: great thanks!
18:49:20 * tbachman notes he has a hard cutoff at noon PDT
18:49:30 <SumitNaiksatam> tbachman: sure, your wish is our command
18:49:37 <tbachman> SumitNaiksatam: lol
18:49:43 <SumitNaiksatam> alrighty, thanks all for joining, and the discussion!
18:49:45 <SumitNaiksatam> bye!
18:49:49 <tbachman> SumitNaiksatam: l8r!
18:49:51 <igordcard> bye
18:49:51 <rkukura> thanks SumitNaiksatam!
18:49:52 <SumitNaiksatam> happy long weekend!
18:49:53 <songole> bye
18:49:54 <rkukura> bye
18:49:57 <ivar-lazzaro> bye!
18:50:02 <SumitNaiksatam> #endmeeting