21:00:05 <strigazi> #startmeeting containers 21:00:06 <openstack> Meeting started Tue Mar 19 21:00:05 2019 UTC and is due to finish in 60 minutes. The chair is strigazi. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:07 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:09 <strigazi> #topic Roll Call 21:00:10 <openstack> The meeting name has been set to 'containers' 21:00:14 <strigazi> o/ 21:00:38 <flwang> o/ 21:00:57 <ttsiouts> o/ 21:04:09 <flwang> strigazi: what's the agenda today? 21:05:32 <strigazi> #topic Announcements 21:05:58 <strigazi> I gave it some mins for anyone else to join. 21:06:31 <strigazi> in a few hours flwang will be the new PTL :) https://governance.openstack.org/election/ 21:06:47 <strigazi> like in two I think 21:06:56 <strigazi> is that correct? 21:07:08 <flwang> strigazi: i think it's true :) 21:07:25 <strigazi> premature congrats :) 21:07:46 <flwang> strigazi: thanks, man. I still need you're around 21:07:59 <ttsiouts> flwang: congrats! 21:07:59 <strigazi> I will be :) 21:08:07 <flwang> i'm mostly like a coordinator of the team 21:08:37 <strigazi> it is indeed 21:08:45 <strigazi> now, for the meeting 21:08:47 <strigazi> #topic Stories/Tasks 21:09:03 <strigazi> I think for resize we are good to merge 21:09:17 <strigazi> resize on update failed works fine for me 21:09:43 <strigazi> maybe we can add later docs that resize works with nova uuid or just the index 21:10:16 <strigazi> for kube-te5lhgvs36ep-minion-5, you can just say nodes_to_remove = ["5"] 21:10:37 <strigazi> +2 21:10:56 <flwang> strigazi: cool, i will add a document to explain the auto scaling, auto healing and includes resize api 21:11:05 <strigazi> just in the reno we need to say that for swarm is not implemented, the nodes_to_remove part 21:11:08 <flwang> it could be a useful user guide 21:11:28 <flwang> strigazi: i will update the releasenote in the api ref patch, is that ok for you? 21:11:45 <strigazi> yeah, no prob. I added +2 21:11:52 <flwang> cheers 21:12:29 <strigazi> for fips, I'm testing the patch: https://review.openstack.org/#/c/641547/ it is a very manual process :( 21:13:11 <colin-> sorry i'm late 21:13:19 <colin-> congrats flwang 21:13:20 <strigazi> colin-: welcome 21:13:24 <flwang> strigazi: what is 'very manual'? 21:13:33 <flwang> colin-: thanks 21:13:33 <colin-> ty strigazi for looking after it this past year (more?) 21:13:36 <strigazi> flwang: all combinations 21:13:47 <strigazi> colin-: almost two :) 21:13:58 <colin-> earned a rest then! 21:14:11 <flwang> strigazi: yep, sometimes flexible means complicated :( 21:14:21 <strigazi> flwang: fips, master_lb, master_lb_fip 21:14:33 <strigazi> colin-: let's see xD 21:15:19 <strigazi> flwang: with ttsiouts we incoreporated the chnage in NGs using conditions and queens as tempalte version 21:16:04 <flwang> strigazi: ah? 21:16:23 <flwang> so we have to upgrade heat to queens? 21:16:39 <strigazi> flwang: we added this label in the WIP templates. 21:16:56 <flwang> label for what? 21:16:58 <strigazi> for the new driver. Not for the existing one 21:17:05 <flwang> ah, i see 21:17:29 <strigazi> this label https://review.openstack.org/#/c/641547/12/doc/source/user/index.rst@387 21:17:45 <strigazi> flwang: you run heat queens, no? 21:18:05 <flwang> strigazi: we're on queens 21:18:15 <strigazi> cool 21:18:56 <flwang> i haven't got time to test lxkong's FIP patch, does the latest PS work for you? 21:19:46 <strigazi> I haven't tested all paths 21:19:59 <strigazi> I'll finish in a bit 21:20:11 <flwang> great, thanks 21:20:19 <flwang> i know it's a tough job 21:20:22 <strigazi> I'll stay a bit late today to finish the upgrade patch too 21:20:36 <strigazi> I can't in the office xD 21:21:20 <strigazi> flwang: any comments for nodegroups? 21:21:31 <strigazi> colin-: fyi: https://review.openstack.org/#/q/status:open+project:openstack/magnum+branch:master+topic:magnum_nodegroups 21:21:43 <flwang> strigazi: i just left a raw comments to suggest use min_node_count and max_node_count 21:22:07 <flwang> so that it's consistent with node_count and i'm using min_node_count and max_node_count for auto scaling and healing 21:22:19 <strigazi> the patches with (zuul +1) are ready. 21:22:26 <strigazi> ok 21:22:28 <flwang> min_nodes is a bit blur 21:22:40 <strigazi> which patchset? 21:22:42 <flwang> strigazi: i haven't got time go through all of them yet 21:22:58 <flwang> the first one, db schema change 21:23:14 <strigazi> flwang: I deployed with an existing cluster (created before NGs) and it works fine. 21:23:29 <strigazi> after the DB migration 21:23:32 <flwang> including all patches? 21:23:48 <strigazi> all current patches that have +1 from zuul 21:23:55 <ttsiouts> flwang: thanks for the review, I will update the PS 21:24:04 <colin-> will take a look at these strigazi thanks 21:24:34 <flwang> ttsiouts: it would be nice if you can add api ref for node groups api 21:24:42 <strigazi> colin-: the actual changes for custom drives shouldn't be a lot 21:24:44 <flwang> and user guide about how to use it 21:24:52 <ttsiouts> flwang: sure, I am on it 21:25:01 <flwang> ttsiouts: great 21:25:34 <ttsiouts> strigazi: should I update the spec to reflect the new names in the fields? e.g s/max_nodes/max_node_count? 21:25:43 <strigazi> yeap 21:25:47 <ttsiouts> cool 21:26:07 <flwang> strigazi: ttsiouts: are you happy with the new name? 21:26:53 <strigazi> flwang: works for me 21:26:58 <ttsiouts> flwang: I think it's better 21:27:06 <ttsiouts> what you proposed 21:28:03 <flwang> cool 21:28:11 <ttsiouts> strigazi, flwang: would an "autoscaled" flag make sense? 21:28:33 <ttsiouts> or should we treat it as a label in each NG? 21:28:37 <flwang> i'm going to use 'auto_scaling_enabled' 21:28:53 <flwang> to be consistent with the other services enabled labels 21:29:04 <flwang> ah, sorry 21:29:20 <flwang> you mean a label/flag against a NG? 21:29:21 <strigazi> flwang: I'm ok with field or label 21:29:31 <strigazi> ttsiouts: ^^ 21:29:58 <ttsiouts> flwang: yes, either a field or a label 21:30:06 <flwang> label is better for some reasons, though it's a bit out of control 21:30:41 <ttsiouts> flwang: in general the nodegroup will inherit labels from the cluster. 21:31:05 <ttsiouts> flwang: overriding them has some problems.. 21:31:10 <flwang> i'm using 'auto_scaling_enabled' https://review.openstack.org/#/c/631378/16/magnum/drivers/common/templates/kubernetes/fragments/enable-auto-healing.sh@209 21:31:56 <flwang> if so, will it cause problems if we have another field for autoscaling 21:32:21 <ttsiouts> flwang: it should be fine I think 21:32:33 <flwang> ttsiouts: so you prefer to have it as a field? 21:33:19 <ttsiouts> flwang: the only thing about labels is that it would be better if we had an operator like "append" to the cluster's labels of sort 21:33:20 <strigazi> I think it would be better but I don't have a strong opinion 21:34:19 <strigazi> (swarm doesn't have it, but probably will never have it) 21:34:29 <ttsiouts> strigazi: yeah.. 21:34:30 <strigazi> it's up to us :) 21:34:56 <strigazi> it doesn't matter for swarm, it can be always false 21:35:01 <ttsiouts> flwang: I don't have strong feelings either 21:35:05 <flwang> ttsiouts: if you do think it's better using 'field', then i'm ok with that 21:35:45 <flwang> ttsiouts: the only thing i'm concerning is the conflicts between auto scaling/healing labels and the field of the NG 21:36:00 <flwang> if you're sure there is no problem, i'm happy 21:36:37 <ttsiouts> flwang: I'll try to check more offline and I'll get back to you on that 21:36:56 <strigazi> with labels the validation is done in the driver, it doesn't make a difference field/label 21:37:08 <flwang> ttsiouts: cool 21:38:21 <strigazi> anything else on this? 21:39:09 <flwang> i'm good, i will do more review, thanks for the great work 21:39:27 <ttsiouts> strigazi: I'm not sure about the update nodegroup api 21:39:56 <ttsiouts> resize makes more sense and it already includes nodegroups... 21:40:20 <strigazi> flwang: ttsiouts I think if the update of NGs doesn't only resize, we can keep resize 21:40:43 <ttsiouts> strigazi: makes sense 21:40:59 <flwang> strigazi: yep, agree, for update, it definitely does more 21:41:40 <strigazi> you mean, makes more sense? 21:41:58 <ttsiouts> strigazi: I meant I agree with you 21:42:03 <ttsiouts> :) 21:42:42 <strigazi> flwang: Do you agree too? 21:43:04 <flwang> keep both update and resize for NG? 21:43:10 <strigazi> no 21:43:28 <strigazi> keep resize, if update doesn't do anything else 21:44:00 <flwang> just like what we have now for cluster update? 21:44:09 <strigazi> yes 21:44:16 <flwang> sure, i agree with that 21:44:20 <strigazi> cool 21:44:32 <ttsiouts> flwang: strigazicool 21:44:47 <strigazi> ttsiouts: anything else for NGs? 21:45:37 <ttsiouts> strigazi: I'll update the PS asap 21:45:44 <strigazi> cool 21:46:45 <strigazi> flwang: ngs (not the new driver necessarily, DB and API) and stein, thoughts? 21:47:33 <flwang> you mean get NG api and db change in stein? 21:47:53 <strigazi> yes 21:48:24 <flwang> i'm ok with that, if the api can throw a nice error 21:48:42 <strigazi> when trying to add a NG? 21:48:53 <flwang> say something like "please go to push strigazi to get it done' 21:49:11 <flwang> personally, i'd like to get NG in stein if we can 21:49:19 <flwang> ttsiouts: ^ are you confident that? 21:49:20 <strigazi> I'm ok with that :) 21:49:30 <colin-> when is our deadline for review in that case? 21:49:34 <colin-> (remind me, plz) 21:49:48 <strigazi> whithin 10 days 21:49:52 <colin-> wow ok 21:49:52 <strigazi> within 10 days 21:49:56 <flwang> https://releases.openstack.org/stein/schedule.html 21:50:13 <ttsiouts> flwang: we need to review the changes 21:50:28 <colin-> assuming that is well underway already/ 21:50:34 <flwang> ttsiouts: yep, sure 21:50:34 <colin-> NG peer review 21:50:51 <strigazi> API and DB are implemented 21:51:01 <flwang> ttsiouts: but we also want to understand if the owner is confidient the work is ready ;) 21:51:03 <strigazi> we can create a set of test tmr to validate 21:51:21 <strigazi> we can create a set of tests tmr to validate 21:51:29 <flwang> the db change looks good to me as long as the name being changed 21:51:38 <flwang> i can approve the db change 21:51:45 <ttsiouts> flwang: cool 21:52:07 <flwang> i will start to review the api change today 21:52:16 <ttsiouts> flwang: thanks! 21:52:38 <flwang> ttsiouts: thank you for your great work 21:52:41 <strigazi> flwang: ttsiouts I think if we prove that, creating a cluster in rocky, do the migration, scale the cluster in stein is pretty good 21:53:13 <strigazi> I have tried this two weeks ago 21:53:15 <flwang> strigazi: we're running out of time 21:53:21 <strigazi> correct 21:53:31 <flwang> for release and this meeting ;) 21:53:47 <flwang> rolling upgrade pls 21:53:50 <strigazi> let's wrap then 21:54:02 <strigazi> I'm staying late offline though 21:54:06 <strigazi> I'll ping you 21:54:16 <flwang> cool, i really really appreciate that 21:54:42 <colin-> ttyl, is the link earlier the best view of outstanding ng reviews? 21:54:48 <colin-> gerrit still confuses me 21:54:50 <strigazi> colin-: thanks for talking a look to ngs 21:54:59 <strigazi> yes 21:55:18 <strigazi> this branch https://review.openstack.org/#/q/topic:magnum_nodegroups+(status:open+OR+status:merged) 21:55:22 <flwang> colin-: you guys comments are more than welcome, pls review it 21:55:35 <colin-> as many as we can think of in the remaining time :) 21:55:54 <colin-> hopefully others have been generating feedback so far? 21:56:26 <flwang> i think now only CC, CERN and Blizzard are around 21:56:44 <strigazi> and brtknr stackHPC :) 21:56:55 <flwang> ah, yes 21:57:08 <flwang> brtknr: apologize 21:58:06 <strigazi> see you around all 21:58:20 <flwang> strigazi: thank you 21:58:27 <strigazi> #endmeeting