Tuesday, 2021-08-31

gthiemongeFYI we're having issues with the active-standby job: https://zuul.openstack.org/builds?job_name=octavia-v2-act-stdby-dsvm-scenario09:30
gthiemongeI opened a story for placement, there's a race condition when octavia wants to launch 2 VMs at the same time09:30
gthiemongewe might have to make the job non-voting if they don't propose a fix quickly09:31
gthiemongeWe also have timeouts in the noop-api jobs, I proposed a workaround to increase the timeout in zuul https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/80655809:33
gthiemongethat's not the right solution, we will have to analyze those timeout issues09:34
opendevreviewGregory Thiemonge proposed openstack/octavia master: Set active-standby job as non-voting  https://review.opendev.org/c/openstack/octavia/+/80674213:16
gthiemonge^ johnsom rm_work 13:17
opendevreviewGregory Thiemonge proposed openstack/octavia master: Add generic network interface management in the amphora  https://review.opendev.org/c/openstack/octavia/+/76119515:21
rm_workHmm... Like nova bug?19:09
johnsomrm_work It's actually in placement19:30
johnsomBut effectively breaks nova if two vms are booted at the same time for the same project.19:30
johnsomhttps://review.opendev.org/c/openstack/placement/+/80673019:32
rm_workfantastic19:32
johnsomYeah, I guess some "consumer_types" stuff merged and broken the database code.19:33
opendevreviewGregory Thiemonge proposed openstack/octavia master: Fix PlugVIPAmphora revert function in amphorav2  https://review.opendev.org/c/openstack/octavia/+/80680420:29
ade_leejohnsom, hey!22:08
johnsomade_lee Hi neighbor. So, it looks like ssh from the fips enabled host to a cirros VM is failing?22:08
ade_leehey -- so yeah -- some context ..22:09
ade_leesome octavia fips tempest tests and a number of other fips tests for other projects are failing because tempest tries to get to the systems using paramiko22:10
ade_leeand paramiko does an md5() ..22:11
ade_leejohnsom, you can see this here -- for instance ..22:11
ade_leewell _ I was going to show you an octavia job but the old results have gone away22:13
ade_leejohnsom, here is a cinder example -- https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_78d/790535/11/check/cinder-tempest-plugin-lvm-lio-barbican-fips/78dd779/testr_results.html22:14
ade_leeanyways - what I was trying to do is replace paramiko with a library that will be fips certified instead in tempest22:15
ade_leejohnsom, and my very wip patch for that is here -- https://review.opendev.org/c/openstack/tempest/+/80627422:15
johnsomOk. Couldn't you also just set a bunch of "disabled_algorithms" for paramiko?22:15
johnsomGotcha, ok, that explains why the non-fips job also failed.22:16
ade_leeyeah22:16
ade_leewhats weird though is that the fips job is failing in a way that is unexpected -- in that when I run the same code locally, it works22:17
ade_leeso something is different in the ci environment22:17
johnsomI have never heard of such a thing... grin22:18
ade_leejohnsom, yeah - so thats why I want to freeze the env for me to poke at22:18
johnsomade_lee Ok, thanks for bringing me up to speed. I have been a bit distracted with the RBAC stuffs and haven't followed what was going on with this patch.22:19
ade_leejohnsom, yeah - I think once we move past that - almost all will likely pass22:20
ade_leepast the paramiko thing that is ..22:20
ade_leeso yeah - its not octavia related (yet), but rather testing tool related.22:21
johnsomFor those playing at home, cirros 0.5.2 has dropbear v2018.76  in it which appears to be incompatible with the libssh change  proposed in tempest.22:56

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!