18:01:03 <SumitNaiksatam> #startmeeting networking_policy 18:01:04 <openstack> Meeting started Thu Aug 6 18:01:03 2015 UTC and is due to finish in 60 minutes. The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:01:08 <openstack> The meeting name has been set to 'networking_policy' 18:01:28 <SumitNaiksatam> #info agenda https://wiki.openstack.org/wiki/Meetings/GroupBasedPolicy#Aug_6th.2C_2015 18:01:53 <SumitNaiksatam> for the past few weeks we have been deliberating as to what is a good time to release Kilo 18:02:10 <SumitNaiksatam> part of the reason in the delay has been that we have users testing this, and providing feedback 18:02:29 <SumitNaiksatam> and we want to make sure that those issues get addressed in the release 18:02:46 <SumitNaiksatam> but given all that, we have most of the things we need 18:03:22 <SumitNaiksatam> for what remains, we can get those committed in the next few days, and/or backport after the release is cut 18:03:42 <SumitNaiksatam> so with that, I would like to propose that we do the release on Aug 14th 18:04:00 <SumitNaiksatam> i already check with most of you, but if you do have concerns with this timeline please do let me know 18:04:16 <SumitNaiksatam> ideally, i will try to do a K-4 release sometime before that 18:04:36 <SumitNaiksatam> thoughts/questions/concerns? 18:05:15 <rkukura> SumitNaiksatam: Is there a list of iterms to get in? 18:05:21 <ransari> SumitNaiksatam - Once Kilo is released, any backport needs to be to both Kilo and Juno 18:05:45 <SumitNaiksatam> ransari: yes, depending on the severity of the issue 18:06:08 <SumitNaiksatam> we will be maintaining Juno and Kilo, while we develop on Liberty 18:06:23 <SumitNaiksatam> rkukura: good question 18:06:34 <SumitNaiksatam> lets discuss that 18:06:37 <SumitNaiksatam> #topic Bugs 18:07:01 <rkukura> SumitNaiksatam: Is my understanding correct that our master branch will continue to require kilo neutron for the short term at least? 18:07:28 <SumitNaiksatam> I am hoping to spend most of the time today on triangulating the list of items to fix before kilo release (as rkukura mentions) 18:07:33 <SumitNaiksatam> rkukura: correct 18:08:14 <ransari> @SumitNaiksatam- couple of bugs that need to get in are: https://review.openstack.org/#/c/197973/, https://review.openstack.org/#/c/198639/, https://bugs.launchpad.net/group-based-policy/+bug/1479706, https://bugs.launchpad.net/group-based-policy-ui/+bug/1482325, 18:08:14 <openstack> Launchpad bug 1479706 in Group Based Policy "GBP Service chaining errors are are not vsible to the user because of not having status attribute" [Undecided,New] 18:08:14 <SumitNaiksatam> rkukura: i am guessing that we will end following a similar process/timeline as to what we did in Kilo (hopefullly we will sync up sooner though) 18:08:15 <openstack> Launchpad bug 1482325 in Group Based Policy UI "Instances in ERROR state don't show in GBP Member list" [Critical,Confirmed] - Assigned to ank (ank.b) 18:08:17 <ransari> and https://bugs.launchpad.net/group-based-policy-ui/+bug/1463928 18:08:17 <openstack> Launchpad bug 1463928 in Group Based Policy UI "External connectivity UI is missing" [Critical,Confirmed] 18:08:45 <SumitNaiksatam> ransari: okay 18:09:16 <SumitNaiksatam> thats a handful! 18:09:38 <SumitNaiksatam> i seem to have completely missed: #link https://bugs.launchpad.net/group-based-policy/+bug/1479706 18:09:38 <openstack> Launchpad bug 1479706 in Group Based Policy "GBP Service chaining errors are are not vsible to the user because of not having status attribute" [Undecided,New] 18:10:01 <rkukura> SumitNaiksatam: What about https://review.openstack.org/#/c/179327/? 18:10:03 <SumitNaiksatam> oh sorry, i think we discussed this 18:10:46 <SumitNaiksatam> i mean we had discussed #link https://bugs.launchpad.net/group-based-policy/+bug/1479706 18:10:46 <openstack> Launchpad bug 1479706 in Group Based Policy "GBP Service chaining errors are are not vsible to the user because of not having status attribute" [Undecided,New] 18:11:00 <SumitNaiksatam> rkukura: lets get to https://review.openstack.org/#/c/179327/ in a minute 18:11:12 <rkukura> ok 18:11:30 <SumitNaiksatam> regarding bug/1479706 i dont think we reached a conclusion 18:11:46 <SumitNaiksatam> but we didnt have magesh that day, is magesh around today? 18:12:07 <magesh> SumitNaiksatam: yes, I am here :- 18:13:07 <SumitNaiksatam> magesh: so my understanding is that we are going to spend some time in Liberty trying to design the asyn interface, while keeping both, sync and async, options available to the user 18:13:36 <SumitNaiksatam> magesh: do you have a short term proposal in mind, that we can later evolve towards the above goal (in the context of the bug you have raised)? 18:14:05 <magesh> SumitNaiksatam: Right, we do need a proper interface to handle the resource status 18:15:10 <rkukura> SumitNaiksatam: I actually meant https://bugs.launchpad.net/group-based-policy/+bug/1432816, which also has schema changes. 18:15:10 <openstack> Launchpad bug 1432816 in Group Based Policy "inconsistent template ownership during chain creation" [High,In progress] - Assigned to Ivar Lazzaro (mmaleckk) 18:15:34 <magesh> SumitNaiksatam: Not having the status impacts the visibility of the errors for the user. I am yet to to work out the details for a short term solution though. 18:15:41 <SumitNaiksatam> magesh: ransari: to confirm, your users needs a fix for this in Kilo, right? (and backported to Juno) 18:15:59 <ransari> yes, 18:16:04 <SumitNaiksatam> okay 18:16:14 <ransari> it impacts the usability fro production deployments 18:16:23 <SumitNaiksatam> ransari: okay 18:17:00 <SumitNaiksatam> magesh: rkukura: can we set aside about 30 minutes Friday morning (US time) to meet in #openstack-gbp and work out a shorter term solution? 18:17:23 <SumitNaiksatam> of course everyone else in the team, who can, should also join 18:17:51 <ransari> ok 18:17:56 <magesh> SumitNaiksatam: I will not be available on Friday morning US time. I will be travelling 18:18:04 <SumitNaiksatam> magesh: ah bummer 18:18:10 <rkukura> SumitNaiksatam: I think so. I don’t see a big problem defining a status attribute for our immediate needs, and adding more values later as things become more asynchronous, provided we don’t break clients. 18:18:11 <SumitNaiksatam> magesh: are you available after this meeting? 18:18:33 <magesh> SumitNaiksatam: okey, Lets discuss after this meeting 18:18:35 <SumitNaiksatam> rkukura: i agree 18:18:45 <SumitNaiksatam> magesh: great 18:18:59 <SumitNaiksatam> rkukura and others hopefully you can stick around a bit longer as well 18:19:09 <ransari> yes 18:19:11 <SumitNaiksatam> so i am just going in the order in which items were posted 18:19:44 <SumitNaiksatam> so #link https://review.openstack.org/#/c/197973 18:19:56 <SumitNaiksatam> “Fix Servicechain Instance update and spec update" 18:20:07 <SumitNaiksatam> ivar-lazzaro: does your objection still stand for this patch? 18:20:30 <ivar-lazzaro> looking 18:21:08 <ivar-lazzaro> ok now I remember this patch 18:21:31 <ivar-lazzaro> I think we had a discussion about finding a common way to handle internal object changes 18:22:06 <ivar-lazzaro> Today we only do it for new PTs in PTGs, we basically added a hook to the NCP that the RMD can call 18:22:53 <ivar-lazzaro> This doesn't scale very well, but I think we can use this method for now until we find a better way 18:23:02 <SumitNaiksatam> ivar-lazzaro: yes i recall we had the discussion 18:23:08 <SumitNaiksatam> ivar-lazzaro: okay 18:23:16 <ivar-lazzaro> it's clearer than calling a "blind" SCI update 18:23:28 <SumitNaiksatam> magesh: can you please respond to ivar-lazzaro’s comment and update accordingly? 18:23:29 <ivar-lazzaro> and finding out later what is going on 18:23:31 <ivar-lazzaro> (IMHO) 18:24:06 <SumitNaiksatam> next one which ransari posted - https://review.openstack.org/#/c/198639/ 18:24:17 <magesh> SumitNaiksatam,ivar-lazzaro : okay, So the idea is to have another method with similar functionality as update SCI, is that correct ? 18:24:44 <SumitNaiksatam> magesh: i agree with ivar-lazzaro that we should make it consistent with existing approach 18:25:21 <SumitNaiksatam> https://review.openstack.org/#/c/198639/ - Gracefully handle parameter updates that affects Service Chain 18:25:32 <magesh> SumitNaiksatam: ok 18:26:07 <SumitNaiksatam> anyone have objections to the above patch/approach ^^^ 18:26:36 <SumitNaiksatam> magesh: we dont need a schema update for this? 18:26:45 <SumitNaiksatam> oh probably not 18:26:51 <magesh> SumitNaiksatam: No, there is no db updates here 18:27:15 <SumitNaiksatam> magesh: and the claim is that this is backward compatible? 18:27:48 <SumitNaiksatam> because we are going from more restrictive to less restrictive 18:27:59 <magesh> SumitNaiksatam: Yes, there is an api update for making classifier ID in SCI an updatable field though 18:28:19 <magesh> the rest are just internal workflow changes 18:28:49 <SumitNaiksatam> magesh: so are we using the classifier ID today in the service chain heat driver? 18:29:22 <magesh> SumitNaiksatam: Yes, In the heat driver we are using it to derive the firewall rule 18:29:38 <SumitNaiksatam> magesh: ah okay, for the firewall rules 18:30:08 <SumitNaiksatam> magesh: however i dont see in our patch that the heat driver is getting updated? 18:30:40 <SumitNaiksatam> i am trying to understand what is being done with the update in your patch 18:30:45 <magesh> SumitNaiksatam: The update scenario is already handled in the heat node driver, although that code is not exercised today 18:30:58 <SumitNaiksatam> magesh: ah okay, i missed that 18:31:08 <magesh> Right now, we delete the chain and recreate it for any updates 18:31:35 <SumitNaiksatam> alright, assuming ivar-lazzaro, rkukura you got a chance to look at this patch, do you foresee any issues with allowing this update? 18:31:36 <magesh> This patch tries to avoid that as much as possible (may be a Policy Rule, PRS, PTG or Classifier updates) 18:32:15 <ransari> right, its not just a classifier update 18:32:46 <SumitNaiksatam> magesh: sounds reasonable to me 18:32:54 <SumitNaiksatam> and the driver can reject the update? 18:32:59 <rkukura> SumitNaiksatam: I haven’t looked closely enough yet to have an opinion 18:33:06 <SumitNaiksatam> rkukura: okay 18:33:26 <magesh> SumitNaiksatam: yes, a driver can reject the update in validate method itself 18:33:35 <SumitNaiksatam> ivar-lazzaro: any objections? 18:34:48 <SumitNaiksatam> ivar-lazzaro is probably reviewing in real time ;-) 18:35:08 <ivar-lazzaro> SumitNaiksatam: multitasking a lot atm ;) 18:35:24 <ivar-lazzaro> I'll review asap, no objections in the concept so far 18:35:28 <SumitNaiksatam> ivar-lazzaro: no worries 18:35:31 <SumitNaiksatam> ivar-lazzaro: okay 18:35:34 <SumitNaiksatam> ransari regading - Launchpad bug 1463928 in Group Based Policy UI "External connectivity UI is missing" 18:35:34 <openstack> Launchpad bug 1463928 in Group Based Policy UI "External connectivity UI is missing" [Critical,Confirmed] https://launchpad.net/bugs/1463928 18:35:36 <ivar-lazzaro> that's obviously better to be graceful when we can 18:35:42 <SumitNaiksatam> ivar-lazzaro: right 18:35:47 <ivar-lazzaro> can't be too rude with the poor chains 18:35:48 <SumitNaiksatam> ransari: this is a big one 18:35:54 <SumitNaiksatam> ivar-lazzaro: lol! 18:36:07 <SumitNaiksatam> ransari: i dont know who is working on this 18:36:50 <SumitNaiksatam> ransari: we also need at least a short discussion on how we want to go about designing this UI before we implement 18:37:59 <ransari> SumitNaiksatam: sorry I stepped away 18:38:05 <SumitNaiksatam> ransari: no worries 18:38:22 <SumitNaiksatam> ransari: i am a bit concerned that we dont have an owner for the above 18:38:24 <ransari> This is an abosulte must for the production deployment 18:38:46 <SumitNaiksatam> ransari: i agree, but we are behind on this 18:38:49 <ransari> Ankaiah can be specified as the owner 18:38:57 <SumitNaiksatam> ransari: okay, good to know 18:39:01 <ransari> But we need to discuss on how to proceed 18:39:06 <SumitNaiksatam> ransari: right 18:39:26 <SumitNaiksatam> ransari: okay lets follow up offline, i would like to know what the user feedback and expectations are on this 18:40:00 <ransari> ok 18:40:01 <SumitNaiksatam> ransari: have we covered all your items? 18:40:42 <SumitNaiksatam> rkukura: you mentioned https://bugs.launchpad.net/group-based-policy/+bug/1432816 18:40:42 <openstack> Launchpad bug 1432816 in Group Based Policy "inconsistent template ownership during chain creation" [High,In progress] - Assigned to Ivar Lazzaro (mmaleckk) 18:41:20 <ransari> yes 18:41:26 <SumitNaiksatam> ransari: ok good 18:41:35 <SumitNaiksatam> so in https://review.openstack.org/#/c/166424/ rkukura you seem to have the last comment 18:41:39 <SumitNaiksatam> but seems like a minor one 18:41:47 <rkukura> SumitNaiksatam: That one is still WIP 18:41:48 <SumitNaiksatam> hate to bug ivar-lazzaro again 18:42:05 <SumitNaiksatam> rkukura: your nova driver depends on the above? 18:42:14 <ivar-lazzaro> yeah the WIP is due to lack of validation so far 18:42:30 <SumitNaiksatam> ivar-lazzaro: okay, seems like rkukura validated, no? 18:42:37 <ransari> Did we cover: https://bugs.launchpad.net/group-based-policy-ui/+bug/1482325 18:42:37 <openstack> Launchpad bug 1482325 in Group Based Policy UI "Instances in ERROR state don't show in GBP Member list" [Critical,Confirmed] - Assigned to ank (ank.b) 18:42:37 <ivar-lazzaro> waiting for rkukura to validate the the VMs are created correctly and the SG rules works 18:42:52 <ivar-lazzaro> SumitNaiksatam: not sure about SG rules being applied 18:43:12 <SumitNaiksatam> ransari: we did not cover, but i will follow up separately, i think it should be pretty doable 18:43:14 <rkukura> Only way I was able to get VMs to use the PT’s port is to use this patch and configure the UUID of the service tenant that owns the VM. Would like the option of specify tenant by name or UUID. 18:43:27 <ransari> SumitNaiksatam: ok 18:43:36 <ivar-lazzaro> rkukura: I think that tenant names could overlap 18:43:43 <ivar-lazzaro> rkukura: that's why UUIDs are used 18:43:44 <rkukura> ivar-lazzaro: Right, looking at the SG rules is at the top of my list 18:44:11 <rkukura> UUIDs are OK, but difficult to configure, especially in devstack 18:44:31 <rkukura> How can different tenants log in if they have the same name? 18:45:28 <ivar-lazzaro> rkukura: good point 18:45:36 <ivar-lazzaro> rkukura: got to investigate this 18:46:01 <rkukura> I should be able to validate the SG rules tomorrow, if not tonight. 18:46:09 <SumitNaiksatam> rkukura: nice 18:46:24 <SumitNaiksatam> rkukura: perhaps you can relay your findings to ivar-lazzaro 18:46:33 <SumitNaiksatam> ivar-lazzaro: perhaps okay to use tenant name? 18:46:40 <rkukura> SumitNaiksatam: I’ll update the review, and email ivar-lazzaro 18:46:45 <SumitNaiksatam> rkukura: thanks? 18:46:56 <ivar-lazzaro> rkukura: thx! 18:47:09 <SumitNaiksatam> ivar-lazzaro: and error out if there is ambiguity/overlap? 18:47:29 <ivar-lazzaro> SumitNaiksatam: sounds good 18:47:33 <SumitNaiksatam> rkukura: sorry i meant - thanks”!” :-) 18:47:51 <SumitNaiksatam> ivar-lazzaro: ok cool 18:48:17 <SumitNaiksatam> so what else did we miss in terms of bugs that are targets for kilo? 18:48:53 <SumitNaiksatam> i will plug the couple of patches that i posted (in response to bugs) - 18:49:04 <SumitNaiksatam> https://review.openstack.org/#/c/209409 (Friendlier consumed/provided PRS input for PTG) 18:49:19 <SumitNaiksatam> https://review.openstack.org/#/c/208200 (Quota support for GBP resources) 18:49:28 <SumitNaiksatam> both are still not completely done 18:49:48 <SumitNaiksatam> there are at least a couple of CLI bugs in addition 18:50:11 <SumitNaiksatam> anything else we missed? 18:50:21 <SumitNaiksatam> rkukura: ivar-lazzaro: magesh: ? 18:50:34 <ransari> SumitNaiksatam: https://review.openstack.org/#/c/209409 definitely improves usability! 18:50:40 <SumitNaiksatam> ransari: okay 18:50:41 <ivar-lazzaro> nope 18:50:48 <SumitNaiksatam> oh we missed ivar-lazzaro’s patch 18:50:55 <SumitNaiksatam> https://review.openstack.org/#/c/208059 18:51:23 <SumitNaiksatam> rkukura: i believe the -1 is mostly very small things? 18:51:53 <SumitNaiksatam> the above is “subnet allocation improvement" 18:51:56 <rkukura> trivial, but didn’t want the bug closed by this patch 18:52:02 <SumitNaiksatam> rkukura: okay 18:52:08 <ivar-lazzaro> ok 18:52:14 <ivar-lazzaro> I'll change the commit message right away 18:52:22 <ivar-lazzaro> and update the rest of the comments later 18:52:25 <SumitNaiksatam> ivar-lazzaro thanks for working on this - we cannot release without this fix 18:52:39 <ivar-lazzaro> I have it as a dependency anyways, so I push new patchsets often 18:52:59 <ivar-lazzaro> SumitNaiksatam: assuming it really improves as much as it should ;) 18:53:04 <rkukura> Might as well address the mispelled variable name too while you are at it. 18:53:46 <SumitNaiksatam> ivar-lazzaro: i think it will, really appreciate you coming up with the interim solution until we leverage the other components 18:54:00 <SumitNaiksatam> ivar-lazzaro: at least it should address most cases 18:54:18 <ivar-lazzaro> hope to see some numbers soon! 18:54:41 <SumitNaiksatam> #topic Testing 18:54:50 <SumitNaiksatam> a quick update on integration gate jon 18:54:53 <SumitNaiksatam> *job 18:55:05 <SumitNaiksatam> we now have a rally job running for both branches 18:55:22 <SumitNaiksatam> i have set the concurrency to 10 18:55:40 <SumitNaiksatam> i do see a few failures at times 18:56:25 <SumitNaiksatam> we havent merged the test in-tree, but if anyone wants to contribute more tests, we can get them in the tree right away (and be able to run it as a part of the external gbp rally package) 18:56:30 <SumitNaiksatam> just reach out to me 18:56:46 <SumitNaiksatam> #topic Packaging update 18:56:49 <SumitNaiksatam> rkukura: anything? 18:56:56 <rkukura> not this week 18:57:00 <SumitNaiksatam> rkukura: okay 18:57:05 <SumitNaiksatam> #topic Open Discussion 18:57:19 <SumitNaiksatam> anything else we missed? 18:57:47 <SumitNaiksatam> would request everyone to please be deligent with the reviews in the next few days 18:58:01 <SumitNaiksatam> it will help our ability to get a good release out 18:58:43 <SumitNaiksatam> rkukura: thanks for fixing the client dependencies! 18:58:50 <SumitNaiksatam> alright thanks everyone! 18:58:56 <rkukura> SumitNaiksatam: I hope I got it right 18:59:06 <SumitNaiksatam> magesh: rkukura ivar-lazzaro ransari, can you join #openstack-gbp ? 18:59:18 <magesh> SumitNaiksatam: ok 18:59:19 <SumitNaiksatam> rkukura: i think it works 18:59:24 <SumitNaiksatam> bye all! 18:59:31 <rkukura> bye 18:59:33 <magesh> bye 18:59:34 <ivar-lazzaro> SumitNaiksatam: ok 18:59:38 <SumitNaiksatam> #endmeeting