Thursday, 2022-01-27

opendevreviewAde Lee proposed openstack/octavia master: Add fips jobs  https://review.opendev.org/c/openstack/octavia/+/79815103:35
noonedeadpunkHey there! I was wondering if there some work/discussions were held regarding cross-AZ scheduling? 08:32
noonedeadpunkAs at this point `availability_zone` seems to be a setting in config, which means you can't create 2 amphoras between AZs08:34
noonedeadpunkor even can't provide nova AZ for LB with API call?08:35
gthiemongenoonedeadpunk: hi08:37
gthiemongethere's an experimental patch by rm_work: https://review.opendev.org/c/openstack/octavia/+/55896208:37
noonedeadpunkoh!08:38
gthiemongeI think he's still using it in his cloud08:38
noonedeadpunkoh, wow08:38
gthiemongeit requires a l2 network shared between the AZs (to handle vrrp traffic)08:38
noonedeadpunkyeah, that's understandable08:38
gthiemongeduring the Yoga PTG, we discussed other approaches (see https://etherpad.opendev.org/p/yoga-ptg-octavia) with BGP announcements08:39
gthiemongebut nobody is working on it08:39
noonedeadpunkFrom notes seems Wallmart was most interested in that?08:41
noonedeadpunkbut ok, that's a good start! We'll what we can do, as we're interested to get this implemented for our usecase.08:42
noonedeadpunk*we'll see08:42
opendevreviewSusanta gautam proposed openstack/octavia master: Add simple notifications for load balancers  https://review.opendev.org/c/openstack/octavia/+/78462808:52
gthiemongenoonedeadpunk: they were interested, but a few years ago08:53
gthiemongeI think it was in 201808:53
noonedeadpunkyeah, based on patch timestamps08:54
noonedeadpunkWell, I'm not promising anything, but we're really interested in this feature right now, and tend to contribute back :)08:55
gthiemongenoonedeadpunk: cool!08:55
noonedeadpunkI was asking not to start re-inveniting bycicle :)08:56
rm_workyeah I worked on rebasing that patch recent;y08:57
rm_workif you want to try to get it back to working, I can help08:58
rm_workjust ping me πŸ™‚ 08:58
rm_workunfortunately I am NOT running that patch anymore but it should be mostly good to go08:58
rm_work(switched companies)08:58
noonedeadpunkwell, things seem to change still last rebase (looking at current conflicts) :)09:04
noonedeadpunkthanks, I won't hesitate then ;)09:05
noonedeadpunkEventually with https://opendev.org/openstack/octavia/commit/815a283823121d563562474eb9048673a95f6e1f things changed for this patch09:08
opendevreviewDmitriy Rabotyagov proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896210:49
opendevreviewDmitriy Rabotyagov proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896211:08
opendevreviewDmitriy Rabotyagov proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896211:09
opendevreviewArx Cruz proposed openstack/octavia-tempest-plugin master: Add scp_connection_disable_sftp option to octavia tests  https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/82663311:19
opendevreviewSusanta gautam proposed openstack/octavia master: Add simple notifications for load balancers  https://review.opendev.org/c/openstack/octavia/+/78462812:13
dav1dbartosHi all, have anyone else expirienced Octavia not finding amphora image on stable/wallaby based API due to "wrong" filter key in glance image driver? It could possibly be just caused by me using stable/wallaby branch with patches from master of Glance. After replacing the "tag" to "tags" in https://opendev.org/openstack/octavia/src/branch/master/octavia/image/drivers/glance_driver.py#L52 it started working.13:23
dav1dbartosI’ve ruled out the possibility of it being just policy mess by checking self.manager.list output and it did not contain the β€žtagβ€œ value, only β€žtagsβ€œ13:23
gthiemongedav1dbartos: I've not heard of that, we have jobs on stable/wallaby that are passing13:28
rm_workyeah I wonder if glance-master changed the key name? but that'd be a pretty big API break for them, I think... they pretty much don't make changes for the last 10 cycles AFAIK lol13:43
rm_worknoonedeadpunk: hmm yeah I thought I already did the rebase for that. I wonder if I didn't push it up >_<13:47
noonedeadpunk:D13:51
opendevreviewDmitriy Rabotyagov proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896214:16
noonedeadpunkrm_work: so with ^ I get only 1 unit test failure - https://paste.opendev.org/show/812402/ I'm not seeing some obvious issue there I believe...14:17
rm_workhmm14:19
rm_workhah yeah i am pretty sure I had worked on this since 2019 πŸ˜€ but must have lost it at some point before I pushed up, RIP14:19
rm_worklooks like maybe you handled most of it tho?14:20
rm_workoh, just tests14:20
rm_worki guess maybe it really was back in 2019 I rebase this? rofl14:26
rm_workcause it definitely has had the Spares Pool code removed14:26
rm_worknoonedeadpunk: i'm not actually sure if this patch needs much work? it's just the test failures you're worried about now I guess?14:31
noonedeadpunkyeah)14:32
noonedeadpunkI just don't see why it fails :)14:32
noonedeadpunkit's smth obvious I'm missing, but I'm jsut not that common with octavia code as for now14:32
noonedeadpunkand it's just one single test that fails now14:32
rm_workdebugging it14:36
rm_workseems like it may be an easy fix14:36
rm_workin `amphora_driver_tasks.py` line 306, when it's looping over the result of the DB fetch...14:37
rm_workthe `db_lb` object is a blank mock, so it has no `db_lb.amphorae` to loop14:37
rm_workthe mocks got out of order14:39
rm_workdid you delete one from the method signature somewhere?14:39
rm_work`mock_loadbalancer_repo_get` is actually `mock_loadbalancer_repo_get_all`14:39
rm_workoh, or amp_repo_get_all14:42
rm_workanyway, it's wrong14:42
rm_workpushing the fix14:42
noonedeadpunkah, yes, I did :(14:46
opendevreviewAdam Harwell proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896214:46
rm_workeasy enough to figure out with a debugger πŸ˜›14:47
noonedeadpunkthanks!14:47
noonedeadpunkah, well, I don't write a lot lately so environemnt is not configured properly atm for pytest14:48
opendevreviewDmitriy Rabotyagov proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896215:44
rm_worklol pep816:06
noonedeadpunkyeah:)16:34
noonedeadpunknow things look green enough16:35
opendevreviewDmitriy Rabotyagov proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896216:43
opendevreviewDmitriy Rabotyagov proposed openstack/octavia master: Experimental multi-az support  https://review.opendev.org/c/openstack/octavia/+/55896216:47
noonedeadpunkno changes there.16:47
noonedeadpunkBut, octavia-v2-dsvm-tls-barbican actually failed with https://zuul.opendev.org/t/openstack/build/ab50ec8fc804480a859ac8125bc31c6d/log/controller/logs/screen-o-cw.txt16:48
noonedeadpunkWhich seems like resulted by https://review.opendev.org/c/openstack/octavia/+/558962/41/octavia/controller/worker/v2/flows/amphora_flows.py16:48
noonedeadpunk`taskflow.exceptions.MissingDependencies: 'linear_flow.Flow: octavia-create-loadbalancer-flow(len=13)' requires ['amphorae'] but no other entity produces said requirements`16:49
opendevreviewGhanshyam proposed openstack/octavia-tempest-plugin master: Add stable/xena jobs on master gate  https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/82520817:37
*** dkehn_ is now known as dkehn18:32
opendevreviewMichael Johnson proposed openstack/python-octaviaclient master: Clarify the client code is not python bindings  https://review.opendev.org/c/openstack/python-octaviaclient/+/82676819:06
opendevreviewGregory Thiemonge proposed openstack/octavia-tempest-plugin master: Add centos-9-stream job  https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/81636919:12
opendevreviewGregory Thiemonge proposed openstack/octavia master: WIP Add support for Centos 9 Stream  https://review.opendev.org/c/openstack/octavia/+/81637019:12
opendevreviewGregory Thiemonge proposed openstack/octavia master: Restart rsyslog from cloud-init  https://review.opendev.org/c/openstack/octavia/+/82676919:12

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!