20:00:04 #startmeeting Octavia 20:00:05 Meeting started Wed Jul 11 20:00:04 2018 UTC and is due to finish in 60 minutes. The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:06 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:08 The meeting name has been set to 'octavia' 20:00:10 Hi folks! 20:00:25 hi 20:00:39 I always worry that I start the meeting in the right channel... lol 20:00:39 o/ 20:00:57 #topic Announcements 20:01:12 Same story, new week, we have a priority bug list for Rocky 20:01:17 #link https://etherpad.openstack.org/p/octavia-priority-reviews 20:01:30 We have made a lot of progress! 20:01:37 But more work to do.... 20:02:14 Also a reminder, python-octaviaclient needs it's final Rocky release by the 26th 20:02:44 And finally: Berlin summit presentation deadline is July 18th 20:03:10 So if you want to submit a talk for Berlin, you have a week to do it. 20:03:18 Any other announcements today? 20:03:55 if you need soccer scores… 20:04:35 Sigh, I am so buried with work right now I haven't even had lunch yet let alone see who is playing... 20:04:49 #topic Brief progress reports / bugs needing review 20:05:19 I have been working on a bunch of stuff. 20:06:11 Fixed the connection limit issue, added a port cleanup that was missing from the provider driver setup, testing UDP patches, 20:06:57 Currently I am focused on internal stuff, but will be starting work on an Active/Standby tempest test. 20:07:50 I started looking at the flavors patch and getting the API reference parts in, but there is more work to do there. I'm not sure it will land for Rocky 20:08:41 Any other updates today? 20:08:53 I have been putting some cycles into privsep but not sure if I will finish it in Rocky 20:09:34 no much from my side. backporting bug fixes to stable branches and some other integration and CI work in TripleO land 20:09:39 Oh, and I wrote a test gate for the migration tool. That was way more work than it's worth, but... So many Ansible bugs/issues 20:09:59 Thanks for staying on top of the backports. 20:10:54 I would like to see this merge so we can start backporting it: https://review.openstack.org/577344 20:11:29 we have to apologize for the asking for a test gate. we probably underestimated the effort required 20:12:10 Oh no worries. it's done now. maybe the code will be useful in the future if we have a need for a "special" gate test. 20:12:54 how can we easily reproduce the issue? 20:13:01 I'm just grumpy about it because it took ~50 patches and a week to make a test run.... 20:13:10 deleting the two amphorae instances would do? 20:13:28 Yes, just nova delete both amps at the same time. 20:13:50 without the patch the LB will go down in a ball of flames 20:14:15 ok. I'll try 20:14:16 with the patch, it will "do the right thing" and repair the LB and both amps 20:15:03 It's a nasty bug that I know people have hit, that is why I want to make sure we get it backported 20:15:37 #topic Versioning patch 20:15:44 #link https://review.openstack.org/559460 20:16:12 I have put up another proposal for versioning our API and communicating that. Please have a look and comment on this approach 20:16:50 As discussed before, I switched the endpoint to be /v2 with an alias for the old /v2.0 path 20:17:52 Any other discussion on that or just review comments? 20:18:15 #topic HMAC hexdigest patch 20:18:22 #link https://review.openstack.org/#/c/571333/ 20:18:40 It has come to my attention that we have a compatibility issue with this change 20:19:01 It has also be backported, so even more sad face 20:19:27 but not released? silver lining? 20:19:37 We have backward compatibility if the control plane is updated, but the amp is an old version. 20:20:12 But, if the amp is new and the control plane is old, the amps get ignored and failed over. 20:20:53 I see a few options here: 20:20:58 1. revert the patches 20:21:15 mmh, I am ok with just documenting - this is an odd case (why would you not update amps with control plane) 20:21:24 2. Add a release note warning the deployer the order things need updating 20:21:58 3. Throw up our hands and run for the hills 20:22:10 * johnsom thinks 3 sounds good about now 20:22:32 CRO:ENG 2:1 20:22:46 Yeah, it is an odd case. I might be ok with adding release notes if you all think that is an ok upgrade requirement 20:22:47 3 sounds exhausting. climbing hills is not easy, but the price we may have to pay 20:23:22 I am fine with (2) 20:23:25 #vote ? 20:24:20 Do we have enough people for that? grin Seems quiet today 20:24:32 what would be an use case where the operator would upload new amp and keep same control plane version? 20:24:53 I can think of 1 or 2 maybe 20:25:27 At least our handy upgrade guide points out the right procedure 20:25:35 #link https://docs.openstack.org/octavia/latest/admin/guides/upgrade.html#cold-upgrade 20:26:29 Yeah, I think it's really people that are grabbing the nightly build image and dropping it into a cloud that doesn't have the updated control plane 20:26:48 The issue will likely go away once we do a stable branch release that includes it 20:27:30 I'd need to think a bit more. maybe it's not something one would face in a normal day 2 using tripleo at least 20:28:01 Ok, I'm leaning towards #2 and adding a release note 20:28:06 +1 20:28:16 Ok, I will do that. 20:28:26 #topic Open Discussion 20:28:27 thanks 20:28:38 That is all I had on the agenda for today. Any other items? 20:29:40 Ok, thanks folks! I'm going to go make lunch... 20:29:54 #endmeeting