| *** gthiemon1e is now known as gthiemonge | 08:23 | |
| opendevreview | Gregory Thiemonge proposed openstack/octavia master: Disable conntrack for TCP flows in the amphora https://review.opendev.org/c/openstack/octavia/+/796608 | 08:47 |
|---|---|---|
| spatel | johnsom hey! morning | 14:27 |
| spatel | what is the good way to restart haproxy without dropping connection? inshort reloading config | 14:27 |
| johnsom | spatel Hello. Yeah, online early for a training session. sigh | 14:27 |
| spatel | no worry! | 14:28 |
| johnsom | spatel This might help: https://www.haproxy.com/blog/truly-seamless-reloads-with-haproxy-no-more-hacks/ | 14:29 |
| johnsom | But, basically yes, you use reload | 14:29 |
| spatel | that doc is very complicated to understand.. | 14:31 |
| spatel | i read and didn't understand if any single liner | 14:32 |
| johnsom | Yeah, it goes into the history behind the zero downtime config reload | 14:32 |
| johnsom | Summary, use the systemd reload command | 14:32 |
| johnsom | http://git.haproxy.org/?p=haproxy-2.4.git;a=blob;f=admin/systemd/haproxy.service.in;h=05fc595796c479b444a804119235388142decab9;hb=HEAD | 14:33 |
| johnsom | Obviously you can't use that file directly, you should use the one provided with your distribution package | 14:34 |
| *** gthiemon1e is now known as gthiemonge | 14:37 | |
| spatel | johnsom that works!!!! | 14:44 |
| spatel | i have 1 million connection so worried :) | 14:44 |
| johnsom | Yeah, just be aware, during the transition you may be using a lot more RAM, so plan accordingly | 14:44 |
| gthiemonge | #startmeeting Octavia | 16:00 |
| opendevmeet | Meeting started Wed Jun 16 16:00:50 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 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 16:00 |
| opendevmeet | The meeting name has been set to 'octavia' | 16:00 |
| gthiemonge | Hi | 16:00 |
| johnsom | o/ | 16:01 |
| haleyb | hi | 16:01 |
| gthiemonge | #topic Announcements | 16:01 |
| gthiemonge | functional tests are broken on octavia master - related to sqlachemy 1.4? | 16:01 |
| gthiemonge | #link https://zuul.opendev.org/t/openstack/build/76fb5298a6b54152bc67722a10e18a23 | 16:02 |
| rm_work | o/ | 16:02 |
| rm_work | Hmm maybe | 16:02 |
| gthiemonge | I'm going to take a look after the meeting, unless someone has an idea | 16:02 |
| rm_work | I put up a patch that I think is also related, with duplicate member creation error handling | 16:02 |
| johnsom | "Deferred loader for attribute 'pool_id' failed to populate correctly" lol | 16:03 |
| rm_work | Where's zzzeek lol | 16:03 |
| gthiemonge | there are a bunch of other errors in the same job | 16:03 |
| gthiemonge | TypeError: unsupported operand type(s) for +: 'ImmutableColumnCollection' and 'list' | 16:03 |
| haleyb | Database health check failed due to: boom.: Exception: boom | 16:04 |
| haleyb | that's a nice exception | 16:04 |
| johnsom | Thanks | 16:05 |
| rm_work | XD | 16:05 |
| rm_work | Yeah that one was probably typed into the test data | 16:05 |
| rm_work | Mocked exception ;) | 16:05 |
| johnsom | I used that for an expected failur | 16:05 |
| rm_work | ^^ yeah in a bunch of places | 16:05 |
| * johnsom is in a parallel meeting, so slightly distracted | 16:06 | |
| gthiemonge | next announcement/topic is: | 16:07 |
| gthiemonge | lower-constraints jobs are broken on python-octaviaclient stable branches (<=victoria) | 16:07 |
| gthiemonge | another one! | 16:07 |
| gthiemonge | #link https://zuul.opendev.org/t/openstack/build/a2ae26e5c9e9442caa04ac5e1addeccd | 16:07 |
| gthiemonge | There are at least 3 stable branches affected: | 16:08 |
| gthiemonge | #link https://review.opendev.org/q/project:openstack%252Fpython-octaviaclient+status:open | 16:08 |
| gthiemonge | and I can see other errors on train | 16:09 |
| haleyb | were we going to drop those jobs from stable branches? i think other components did | 16:09 |
| haleyb | oh, we did in octavia too | 16:10 |
| gthiemonge | Yeah we can drop the lower-constraints job on stable branches, I think we already did it in the past for similar reason | 16:10 |
| johnsom | No. we decided not to as our tests, unlike some other projects, actually test lower constraints | 16:10 |
| johnsom | Yeah, I think we agreed that stable could go | 16:10 |
| gthiemonge | ok | 16:11 |
| gthiemonge | I'll take a look at those patches as well | 16:11 |
| haleyb | johnsom: so was that a yes or a no? | 16:11 |
| rm_work | yes for stable-branches, no for master branch | 16:12 |
| rm_work | AFAIU | 16:12 |
| johnsom | Yes for stable. No we are not going to remove it for the main branch | 16:12 |
| johnsom | Right | 16:12 |
| gthiemonge | rm_work: Yes, that was our agreement from a previous meeting | 16:12 |
| rm_work | then we're all agreed! again! :D | 16:12 |
| haleyb | agreed, can do like we did for octavia repo then | 16:13 |
| gthiemonge | yep | 16:13 |
| gthiemonge | great! | 16:13 |
| gthiemonge | any other announcements? | 16:13 |
| gthiemonge | ok | 16:14 |
| gthiemonge | #topic Brief progress reports / bugs needing review | 16:15 |
| gthiemonge | not such much upstream work for this week, I've been focusing on downstream tasks. | 16:15 |
| gthiemonge | but I proposed a commit to disable conntrack for TCP connections in the amphora | 16:16 |
| gthiemonge | #link https://review.opendev.org/c/openstack/octavia/+/796608 | 16:16 |
| johnsom | Yeah, all I have is the client patch for large cloud improvements and the associated backports | 16:16 |
| gthiemonge | conntrack is always enabled in Octavia but we should use it only for LVS-based listeners (UDP/SCTP) | 16:17 |
| johnsom | +1 to that | 16:17 |
| johnsom | Wish we could just remove that kernel module again, but at least we can bypass it for non-UDP based traffic | 16:17 |
| gthiemonge | I heard people complaining about the kernel messages that are logged when the conntrack table is full, and then an haproxy with a high maxconn was killed by the oom-killer. | 16:17 |
| rm_work | bleh :/ | 16:18 |
| rm_work | oh, so probably (maybe?) related to SQLAlchemy upgrade: https://review.opendev.org/c/openstack/octavia/+/795637 | 16:18 |
| rm_work | i should update the title to mention it's also for listeners, pre-emptively | 16:18 |
| gthiemonge | #topic Open Discussion | 16:22 |
| gthiemonge | Any other topics today? | 16:22 |
| haleyb | can we decide on an irc channel? :-p | 16:22 |
| haleyb | https://review.opendev.org/c/openstack/project-config/+/793999 | 16:22 |
| gthiemonge | johnsom: rm_work: should we propose a pytohn-octaviaclient release for Xena? | 16:22 |
| haleyb | that way i can update the contributor doc | 16:23 |
| gthiemonge | haleyb: yeah sorry I was a bit lazy on that one | 16:23 |
| johnsom | Yes, if it's not broken. grin | 16:23 |
| gthiemonge | I'll fix the commit and re-propose it | 16:23 |
| gthiemonge | johnsom: master is broken? | 16:23 |
| gthiemonge | johnsom: I'll check what you did :D | 16:23 |
| johnsom | I don't know | 16:23 |
| opendevreview | Adam Harwell proposed openstack/octavia master: Make duplicate constraint checks more generic https://review.opendev.org/c/openstack/octavia/+/795637 | 16:24 |
| gthiemonge | so the next meeting will probably be held in openstack-octavia | 16:25 |
| rm_work | I wish we could do channel redirects... ... can we do that? | 16:25 |
| gthiemonge | if my commit passes the CI :D | 16:25 |
| rm_work | I still don't particularly like switching the channel name | 16:25 |
| gthiemonge | rm_work: I don't think OFTC supports it | 16:26 |
| johnsom | Right, I don't think OFTC supports it | 16:26 |
| rm_work | i think it's 50/50, maybe some people are looking for the "openstack-octavia" channel and can't find it? but I think once we switch it's just going to be some people failing to find the lbaas channel <_< | 16:26 |
| gthiemonge | we will be in the lbaas channel, redirecting people to octavia ;-) | 16:26 |
| haleyb | should i flip a coin? | 16:26 |
| rm_work | my followup for that is, if it's a coin flip, why bother changing | 16:27 |
| rm_work | we have like 8 years of people knowing where this channel is :D | 16:27 |
| rm_work | and docs showing it is here, etc | 16:27 |
| haleyb | ok, thumb war then | 16:27 |
| rm_work | lol | 16:27 |
| gthiemonge | Yeah you have a point here | 16:28 |
| rm_work | i mean whatever, if you're set on changing the name, i guess that's fine... it just seems pointless to me | 16:28 |
| rm_work | go sit in #openstack-octavia and set the topic to "head over to #openstack-lbaas" | 16:29 |
| rm_work | ;) | 16:29 |
| johnsom | We can easily make openstack-octavia disappear as well. | 16:29 |
| haleyb | i proposed it just as we were moving, pendulum never swung far enough i guess | 16:29 |
| rm_work | i mean there's two ways to look at that -- we're moving, so lets rip the bandaid and do all the change at once; or: we're already doing one change, lets upset things as little as possible | 16:30 |
| rm_work | but wasting too much time discussing this is just bikeshedding :D | 16:30 |
| rm_work | I'm +0 | 16:31 |
| rm_work | so if the change is merging, i'll see you over there :P | 16:31 |
| gthiemonge | ok, I'll abandon the change, we will stay on #openstack-lbaas | 16:31 |
| gthiemonge | change is still WIP-1 and 2xCR-1 | 16:32 |
| gthiemonge | so that would be extra work for me | 16:32 |
| haleyb | ok, i'll update by contrib doc patch then to point to OFTC | 16:32 |
| gthiemonge | haleyb: thanks, rm_work will review that change | 16:32 |
| rm_work | I will :P | 16:32 |
| rm_work | ping me! | 16:32 |
| haleyb | oh, and while we're on the simple patch wagon, https://review.opendev.org/c/openstack/python-octaviaclient/+/795902 fits the bill | 16:33 |
| gthiemonge | haleyb: I'll review it thanks | 16:34 |
| gthiemonge | Ok guys, anything else? | 16:34 |
| gthiemonge | Ok thank you! | 16:36 |
| gthiemonge | #endmeeting | 16:36 |
| opendevmeet | Meeting ended Wed Jun 16 16:36:43 2021 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 16:36 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/octavia/2021/octavia.2021-06-16-16.00.html | 16:36 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/octavia/2021/octavia.2021-06-16-16.00.txt | 16:36 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/octavia/2021/octavia.2021-06-16-16.00.log.html | 16:36 |
| opendevreview | Brian Haley proposed openstack/octavia master: Update contributor documentation https://review.opendev.org/c/openstack/octavia/+/793216 | 16:36 |
| haleyb | rm_work: there you go ^^ | 16:37 |
| johnsom | Thanks Greg | 16:37 |
| rm_work | +2 | 16:38 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!