Wednesday, 2021-05-05

*** sapd1_x has quit IRC00:28
openstackgerritAdam Harwell proposed openstack/octavia master: Use vip_subnet_selection_tag to select subnets
*** njohnston has quit IRC01:03
*** tkajinam has joined #openstack-lbaas01:13
openstackgerritTakashi Kajinami proposed openstack/octavia-dashboard master: Support policy-in-code and deprecated policy
*** rcernin has quit IRC03:05
*** rcernin has joined #openstack-lbaas03:16
openstackgerritAdam Harwell proposed openstack/octavia master: Use vip_subnet_selection_tag to select subnets
*** psachin has joined #openstack-lbaas03:37
*** sapd1_x has joined #openstack-lbaas04:15
*** vishalmanchanda has joined #openstack-lbaas04:51
*** xgerman has quit IRC05:03
*** sapd1_x has quit IRC05:30
*** damien_r has joined #openstack-lbaas06:27
*** damien_r has quit IRC06:31
*** damien_r has joined #openstack-lbaas06:37
openstackgerritGregory Thiemonge proposed openstack/octavia-tempest-plugin master: Move ubuntu jobs to focal
*** rpittau|afk is now known as rpittau07:12
*** sapd1_x has joined #openstack-lbaas07:18
*** rcernin has quit IRC07:21
*** rcernin has joined #openstack-lbaas07:25
*** sapd1_x has quit IRC07:42
*** sapd1_x has joined #openstack-lbaas07:49
*** rcernin has quit IRC07:59
*** ianychoi_ has quit IRC08:13
*** rcernin has joined #openstack-lbaas08:23
*** rcernin has quit IRC08:29
*** QG has joined #openstack-lbaas08:37
*** rcernin has joined #openstack-lbaas08:38
*** rcernin has quit IRC08:42
*** rcernin has joined #openstack-lbaas08:46
*** rcernin has quit IRC09:30
*** osmanlicilegi has quit IRC09:50
*** rcernin has joined #openstack-lbaas09:54
*** rcernin has quit IRC10:05
*** osmanlicilegi has joined #openstack-lbaas10:15
*** rcernin has joined #openstack-lbaas10:52
*** rcernin has quit IRC10:58
*** rcernin has joined #openstack-lbaas11:02
*** rcernin has quit IRC11:06
*** sapd1_x has quit IRC11:18
*** rcernin has joined #openstack-lbaas11:24
*** rcernin has quit IRC11:28
*** rcernin has joined #openstack-lbaas11:45
*** rcernin has quit IRC11:55
*** njohnston has joined #openstack-lbaas12:03
openstackgerritAdam Harwell proposed openstack/octavia master: Use vip_subnet_selection_tag to select subnets
*** rcernin has joined #openstack-lbaas12:18
*** rcernin has quit IRC12:23
openstackgerritMerged openstack/octavia master: Optimize CountPoolChildrenForQuota task in amphorav2
openstackgerritMerged openstack/octavia master: Spare pool removal
openstackgerritMerged openstack/octavia master: Fix comment for the ca_certificates_file opt
openstackgerritArieh Maron proposed openstack/octavia-tempest-plugin master: Updating _test_pool_CRUD to enable testing of updates to the load balancer algorithm:
*** dulek has quit IRC13:22
*** gregraka has quit IRC13:25
*** dulek has joined #openstack-lbaas13:35
openstackgerritArieh Maron proposed openstack/octavia-tempest-plugin master: Updating _test_pool_CRUD to enable testing of updates to the load balancer algorithm:
rm_workhmm, now that the spares pool code is gone... I wonder what happens to amps that existed in the spares pool, if it was configured? :D do they just sit there I guess and you have to use the amp delete command to admin-remove them? :P14:06
rm_workwill that work properly?14:06
* rm_work will have to poke at that later14:06
rm_workdidn't realize you were so far along that the removal was already merging! great to see it gone, very good simplification. just my paranoia at work :P14:07
gthiemongerm_work: the old spare amps are detected as unattached/bogus amps by the HM, and then HM deletes them14:08
rm_workWill have to read through the rest of that patch later14:09
*** gregraka has joined #openstack-lbaas14:20
johnsomrm_work is also an option14:22
rm_workYeah that's what I was referring to14:25
rm_workBut looks like Gregory handled it :)14:26
*** rpittau is now known as rpittau|afk15:52
gthiemonge#startmeeting Octavia16:00
openstackMeeting started Wed May  5 16:00:20 2021 UTC and is due to finish in 60 minutes.  The chair is gthiemonge. Information about MeetBot at
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.16:00
*** openstack changes topic to " (Meeting topic: Octavia)"16:00
openstackThe meeting name has been set to 'octavia'16:00
gthiemongeHi folks16:00
gthiemongeOops, it looks like I forgot to update the agenda :/16:00
gthiemonge#topic Announcements16:01
*** openstack changes topic to "Announcements (Meeting topic: Octavia)"16:01
gthiemongeCall for community goals16:02
gthiemongeThere's call for new community goals for the Y release16:02
openstackgerritArieh Maron proposed openstack/octavia-tempest-plugin master: Updating _test_pool_CRUD to enable testing of updates to the load balancer algorithm:
gthiemongejohnsom: are there community goals for X? I cannot find anything16:03
johnsomNo, there were not16:03
johnsom<meme of "vote for none of the above" pops in my head>16:04
johnsomThat is the list of "selected goals"16:04
haleybWe took cool-down cycle goal step for Xena16:04
haleybcycle [7], so no selected goals for Xena.16:04
gthiemongeyeah there's a 404 ->
johnsomI see there is a strong list of proposals:16:05
* johnsom votes for "placeholder"16:05
gthiemongenext announcement:16:06
gthiemongeDevstack dropping support for Ubuntu Bionic (on master)16:06
gthiemongeWe received an email on openstack-discuss16:07
gthiemongebecause we still have Ubuntu Bionic jobs on our master branch.16:07
gthiemongeDevstack is going to drop the Bionic support and asks people to move to Focal.16:07
johnsomYeah, I think this was delayed due to the "switch base jobs to centos" discussion16:07
gthiemongeThere's an open review for octavia-tempest-plugin:16:07
gthiemongejohnsom: oh cool, but switching to focal doesn't look bad16:08
gthiemongeNote that train and ussuri branch will still use Bionic16:08
johnsomYeah, the plan for centos never materialized. I think the proposed patch is still broken.16:09
johnsomSo, moving forward with focal is a good idea16:09
gthiemongeWe also need to send a patch for the octavia, there are some Bionic jobs:16:09
gthiemongeAny other announcements?16:10
gthiemonge#topic Brief progress reports / bugs needing review16:11
*** openstack changes topic to "Brief progress reports / bugs needing review (Meeting topic: Octavia)"16:11
gthiemongeI'm working on the amphorav2, I have an new patch that needs reviews:16:12
gthiemongeThere's still an open bug/review in taskflow, but I think the amphorav2 jobs are now in a good shape.16:13
johnsomI have mostly been doing reviews. I did update the focal nightly build patch:16:13
johnsomgthiemonge I answered your question on that16:13
johnsomI think I also re-reviewed the taskflow patch this morning16:14
gthiemongeI hope we will be able to make amphorav2 the default driver soon16:14
gthiemonge#topic Open Discussion16:16
*** openstack changes topic to "Open Discussion (Meeting topic: Octavia)"16:16
QGI just want to introduce myself, I work at OVHCloud and we have started to integrate Octavia, if we can help, make tests, or reviews16:18
johnsomHi QG, welcome!16:18
johnsomWe certainly can use more reviewers, so please feel welcome to help us review patches.16:18
gthiemongeQG: Welcome!16:18
johnsomIt's a great way to learn the code too.16:18
QG Thank you16:19
johnsomAlso, if you have questions, this channel is pretty good about helping folks out.16:19
QGjohnsom Yes, thanks16:20
gthiemongeSomeone sent me this link to your site16:20
QGgthiemonge  That's exactly it16:20
gthiemongethat explains the RFE you opened for the req/s limitation16:21
johnsomAh, yes. I had some questions/comments on that RFE16:22
QGAs we deploy Public Cloud we need to a way to do some req/s limitation16:24
johnsomYou don't just charge by bytes in/out? grin16:24
QGNo, we don't charge for the network16:25
johnsomYeah, we have a few features along that line today, but the RFE seemed to be a new request.16:25
johnsomDo you have a developer that will work on that feature?16:25
QGYeah Pierre-Yves, who opened the RFE will work on it16:26
johnsomIt wasn't on the PTG topic list, so wasn't something we discussed. I think we can work out the details in the RFE story.16:26
QGyes we missed the PTG by a week16:27
johnsomOk cool. Please let them know they can ask questions here and there are comments on the RFE.16:27
johnsomWe are happy to help them be successful!16:27
QGThanks ! sure no problem16:27
gthiemongeHey, I'm on PTO next wednesday, should we cancel the meeting?16:28
johnsomAh, good question. If we want to have it I can volunteer to run it. However, if we want a break I am good with that too16:29
gthiemongeOk we can skip the next meeting16:30
haleybi'm fine with canceling16:30
gthiemongeI'll send an email on openstack-discuss16:30
gthiemongeany other topics today?16:31
gthiemongeThanks everyone16:32
johnsomThanks gthiemonge16:33
*** openstack changes topic to "Discussions for OpenStack Octavia | Priority bug review list:"16:33
openstackMeeting ended Wed May  5 16:33:03 2021 UTC.  Information about MeetBot at . (v 0.1.4)16:33
openstackMinutes (text):
*** psachin has quit IRC16:36
*** QG has quit IRC16:37
*** zzzeek has joined #openstack-lbaas16:45
*** zzzeek has quit IRC16:50
*** zzzeek has joined #openstack-lbaas16:52
*** osmanlicilegi has quit IRC16:54
*** osmanlicilegi has joined #openstack-lbaas17:00
*** zzzeek has quit IRC17:04
*** zzzeek has joined #openstack-lbaas17:08
*** vishalmanchanda has quit IRC18:46
openstackgerritMerged openstack/octavia master: Add nightly test image build job for focal
*** xgerman has joined #openstack-lbaas22:02
*** armax has joined #openstack-lbaas22:58
*** armax has left #openstack-lbaas22:58
*** rcernin has joined #openstack-lbaas23:04
rm_workFINALLY I think I see what's wrong23:21
openstackgerritAdam Harwell proposed openstack/octavia master: Use vip_subnet_selection_tag to select subnets
rm_worknetwork driver isn't a singleton, and with the no-op driver where state is just stored internally, this does not work <_<23:21
rm_workmaybe the correct fix is just to make the `utils.get_network_driver()` function a singleton handler?23:22
rm_workjohnsom: can you think of any reason we don't do this for the network driver?23:22
rm_workit takes no context or anything for instantiation...23:23
johnsomUgh, I remember there was a discussion of why the network driver isn't a singleton, but ... I don't remember any details.23:23
rm_workI'm gonna try making the change and see if it explodes stuff23:23
johnsomToken lifetime?23:23
rm_workyeah that could be it23:23
rm_workdoes it not have a re-auth built in to the SDK?23:24
rm_workhmmm or would we be circumventing that basically23:24
rm_workyeah reading through it23:25
johnsomYou are in an area of the code I have only put duct tape on, I wasn't involved in it's origination much.23:25
rm_workbut can a default KeystoneSession not re-auth?23:25
* haleyb mic drops with the OVN provider network client link,
haleybthat uses Singleton23:25
rm_workso are you saying we should avoid doing that, because OVN does it? :P23:26
johnsomSounds right to me23:26
rm_workcouldn't avoid that dig :P23:26
haleyball i know is someone fixed that code, and i noticed Octavia was similar and didn't know who was right23:27
haleyblet sleeping dogs lie23:27
rm_workthis is way more fancy than I was going to do for a singleton <_<23:27
johnsomYeah, in the heady days of "we will support multiple network implementations" this decision was a bit higher up than the neutron client level23:27
rm_workI was just gonna make a global and set it :P23:27
rm_workis there an advantage to being so fancy with metaclassing23:28
johnsomSo, wait, that client already is a singleton for the neutron auth part.23:28
haleybrm_work: that would probably work too, i only know we saw memory (?) issues somewhere and doing it like that fixed it23:28
rm_workoh, because keystone is?23:28
johnsomIt's really down to the network driver code itself, which is a lot safer23:28
johnsomUgh, why did you distract me from this deep code review on designate... sigh23:30
rm_workso actually it's up to the specific driver?23:30
rm_workso I could make the no-op driver specifically be a singleton23:30
johnsomYes, that is true for sure. (thought it was really)23:30
haleybjohnsom: yes, i think when i saw that line i went hmmm as well, should be sufficient in OVN as well, or maybe this all belons in octavia-lib :)23:30
johnsomwhy wouldn't the no-op be a singleton?????23:30
johnsomhaleyb I think it's a fail if drivers go through octavia-lib to go through neutron client IMO.23:31
rm_workright!? it should be23:31
rm_workbut it doesn't appear to be23:31
johnsomrm_work I see zero risk in doing that in no-op23:32
haleybjohnsom: well, we can define the same thing in many places then, and do it differently, or just once23:32
johnsomIf anything, you should use neutron-lib.... lol23:33
haleybnow *that's* a zinger!23:33
openstackgerritAdam Harwell proposed openstack/octavia master: Use vip_subnet_selection_tag to select subnets
openstackgerritAdam Harwell proposed openstack/octavia master: Use vip_subnet_selection_tag to select subnets
rm_worknm the Singleton stuff is in oslo_service ^^ :P thanks for the find johnsom23:47
* johnsom puts on his oslo liaison badge 23:49

Generated by 2.17.2 by Marius Gedminas - find it at!