16:00:36 <gthiemonge> #startmeeting Octavia
16:00:36 <opendevmeet> Meeting started Wed Jun 23 16:00:36 2021 UTC and is due to finish in 60 minutes.  The chair is gthiemonge. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:36 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:36 <opendevmeet> The meeting name has been set to 'octavia'
16:00:43 <gthiemonge> Hi everyone!
16:01:16 <haleyb> hi
16:02:04 <gthiemonge> where is johnsom?
16:03:07 <gthiemonge> #topic Announcements
16:03:14 <gthiemonge> Gates are green!
16:03:24 <gthiemonge> Thanks to the reviewers!
16:04:12 <gthiemonge> (we had lower-constraints issues on python-octaviaclient stable branches, and some bugs with sqlalchemy 1.4 on octavia master)
16:04:47 <gthiemonge> and I don't have any other announcements
16:05:48 <gthiemonge> #topic Brief progress reports / bugs needing review
16:06:02 <gthiemonge> I've been working on fixing some non-voting jobs
16:06:18 <gthiemonge> I switch the centos 8 job to centos stream 8
16:06:25 <gthiemonge> "switched"
16:06:46 <gthiemonge> #link https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/698450
16:06:53 <gthiemonge> (based on a commit from cgoncalves)
16:07:30 <gthiemonge> it depends on a devstack patch/hack that fixes dib when it is deployed in a devstack env
16:07:39 <gthiemonge> I hope it will be merged soon
16:07:49 <gthiemonge> Then the goal is to set it as a voting job
16:08:17 <gthiemonge> and perhaps we will be able to switch the jobs on master branch from ubuntu to centos
16:08:43 <gthiemonge> I also worked on the two-node job
16:09:13 <gthiemonge> the switch to ML2/OVN broke the patch, I'm still fighting with FIPs in the scenario tests
16:09:18 <gthiemonge> #link https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/773888
16:09:31 <gthiemonge> haleyb: maybe I will need your expertise here
16:09:35 <haleyb> gthiemonge: switch or add a centos?  should we not keep the ubuntu config?
16:09:47 <haleyb> gthiemonge: ack
16:10:58 <gthiemonge> haleyb: we will keep the ubuntu config. basically, octavia-v2-dsvm-scenario will be centos based on master, and we will have a octavia-v2-dsvm-scenario-ubuntu
16:11:10 <haleyb> +1
16:11:25 <gthiemonge> config for stable branches will be untouched
16:11:51 <gthiemonge> (this is also cgoncalves' work)
16:12:04 <gthiemonge> #link https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/746996
16:14:21 <gthiemonge> #topic Open Discussion
16:17:00 <haleyb> gthiemonge: question on your generic network interface patch, https://review.opendev.org/c/openstack/octavia/+/761195
16:17:15 <haleyb> i went to review again but looks like you were going to update it?
16:17:34 <gthiemonge> hmmm
16:17:40 <gthiemonge> I don't remember :D
16:17:49 <haleyb> all my nitty comments
16:18:11 <opendevreview> Michael Johnson proposed openstack/octavia-tempest-plugin master: Remove unused dead code  https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/797715
16:18:13 <haleyb> if it's just updating things to use a different IPv6 prefix i wouldn't bother
16:18:19 <gthiemonge> Oh yeah you're right
16:18:37 <haleyb> i'v6 got ipv6 on the brain this morning
16:19:07 <gthiemonge> I'll try to update it before the end of the week
16:19:25 <gthiemonge> I hope the CI won't be unhappy
16:19:35 <haleyb> a shout-out on reviews for that one
16:19:51 <haleyb> shh, don't talk about the CI
16:19:55 <johnsom> o/
16:20:12 <gthiemonge> I think I also need to update it to fix some issues with the host_routes
16:20:15 <johnsom> Sorry, parallel discussions
16:21:51 <gthiemonge> haleyb: thanks for the heads up, because I forgot to send the last patchset for review, I almost lost it
16:22:02 <johnsom> I had some comments no the generic network patch, but got interrupted reviewing.
16:22:13 <johnsom> I can post those (if I didn't), but I'm not done yet
16:22:40 <gthiemonge> ok np
16:23:01 <johnsom> Ha, yeah, host routes is the last comment on my local notes....
16:23:09 <johnsom> So, I guess that was in fact missing.
16:24:20 <johnsom> So, did I see above that we are switching the main tests to use centos? Are we ready for that? The centos gate hasn't passed in over a year as far as I am aware
16:24:47 <gthiemonge> not yet
16:24:55 <johnsom> Ok
16:25:04 <gthiemonge> The (my) plan is to restore the centos-stream-8 job
16:25:10 <johnsom> Yeah, I would like to see it pass for at least a bit before we start proposing that switch.
16:25:21 <gthiemonge> when we consider that it is stable, we can switch it to voting
16:25:21 <johnsom> Ok, yeah, that sounds like a good idea
16:25:41 <johnsom> It would be super awesome to have a working and voting centos job again
16:26:17 <gthiemonge> I agree
16:27:49 <johnsom> A different topic, I just posted a patch based on a discussion in the qa channel. We may have some legacy/dead code in the tempest plugin.
16:30:06 <gthiemonge> ok, it passes the CI, I'll CR+2 it ;-)
16:30:12 <gthiemonge> "if'
16:31:02 <gthiemonge> thanks johnsom
16:31:20 <haleyb> johnsom: pep8 failure :(
16:31:54 <johnsom> This may need a patch or two, we want to see if it's truely dead or not.
16:32:04 <johnsom> pep8 for a one line delete???? hmmm lol
16:32:51 <johnsom> Ah, no longer needed import.
16:32:58 <haleyb> bingo
16:33:04 <johnsom> Ok, I want to let the tempest run, then I will fix that
16:33:12 <gthiemonge> ack
16:33:17 <johnsom> Take a day off and you get rusty. grin
16:34:06 <gthiemonge> One last thing about the tempest tests, I noticed that one test failed randomly in a CI job (ipv6 test with listener_with_allowed_cidrs)
16:34:13 <gthiemonge> I proposed a fix:
16:34:20 <gthiemonge> #link https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/797387
16:34:46 <gthiemonge> it would improve the stability of our jobs
16:35:33 <gthiemonge> sadly, we are mixing the ipv6 address notations in the tempest test ([aaaa:bbbb::1] va aaaa:bbbb::1)
16:35:37 <gthiemonge> and the fix is not pretty
16:36:37 <haleyb> gthiemonge: should it always use brackets for IPv6?
16:36:49 <gthiemonge> it depends on the function :D
16:37:52 <gthiemonge> make_request expects brackets, check_members_balanced doesn't :/
16:37:57 <haleyb> hmmm, does it break UDP ?
16:38:20 <gthiemonge> make_requset expects brackets for HTTP/TCP
16:38:56 <haleyb> oh, i expanded below and see the other occurrence
16:41:19 <gthiemonge> maybe we can clean that stuff in an other commit
16:41:59 <johnsom> Yeah, I want to look at that. I know we have accommodations for that somewhere, maybe just not in the right place.
16:42:20 <haleyb> right, callers should pass the correct address to check_members_balanced()
16:42:29 <haleyb> follow-on is good
16:43:29 <haleyb> or make_request() can do the same hack
16:44:05 <gthiemonge> if the brackets are not already there
16:44:24 <johnsom> Yeah, maybe those validators should do a little AI and figure out if the brackets are there already, or are needed.
16:44:38 <gthiemonge> lol
16:45:29 <gthiemonge> we can discuss it in the review ;-)
16:45:40 <gthiemonge> any other topics for this week folks?
16:47:08 <gthiemonge> Ok
16:47:13 <gthiemonge> Thanks everyone!
16:47:16 <gthiemonge> #endmeeting