16:01:28 #startmeeting containers 16:01:33 Meeting started Tue Mar 7 16:01:28 2017 UTC and is due to finish in 60 minutes. The chair is adrian_otto. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:34 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:37 The meeting name has been set to 'containers' 16:01:37 #link https://wiki.openstack.org/wiki/Meetings/Containers#Agenda_for_2017-03-07_1600_UTC Our Agenda 16:01:41 #topic Roll Call 16:01:45 Adrian Otto 16:01:46 o/ 16:01:48 Spyros Trigazis 16:01:51 o/ 16:01:54 Jaycen Grant 16:02:00 Corey O'Brien 16:02:00 o/ 16:02:03 o/ 16:02:27 hello Drago strigazi hieulq_ jvgrant_ coreyob yatinkarel and vijendar 16:03:05 #topic Announcements 16:03:33 we do not have prepared announcements for today. Any announcements from team members? 16:03:59 gate-experimental-functional-dsvm-magnum-k8s-multinode-ubuntu-xenial I'm waiting for this :) 16:04:08 multinode gate 16:04:19 when is that expected, strigazi ? 16:04:34 It's merged and a job is running as we speak 16:04:48 how exciting!! 16:04:49 I hope it will improve our ci 16:04:58 nice! 16:05:03 indeed. 16:05:08 strigazi, So with multinode cluster loadbalancer 16:05:19 it's election day (again) in the US 16:05:37 at least it is in California, for local elections 16:05:52 ok, let's move on 16:05:55 we might be able to test lbaas, but let's wait 16:06:02 strigazi, Ok 16:06:19 #topic Review Action Items 16:06:19 (none) 16:06:31 #topic Blueprints/Bugs/Reviews/Ideas 16:06:35 Essential Blueprints 16:06:41 #link https://blueprints.launchpad.net/magnum/+spec/flatten-attributes Flatten Attributes [strigazi] 16:07:27 As I was moving forward with this I realized that we move all the fields from CT to cluster. The problem is we don't want all there fields 16:07:37 eg apiport 16:07:59 the Driver resource will have a blob 16:08:24 and cluster will have a bunch of attrs that not all drivers need 16:08:28 right 16:08:42 Should we procceed this way? 16:08:57 thoughts? 16:09:04 we should first flatten what we have into Cluster 16:09:28 adrian_otto: +1, i think we should do this in steps. 16:09:30 then reduce that list as a secondary effort with the driver config 16:09:37 ok 16:09:41 The alternative would be to implement driver config first 16:09:42 try not to conflate too many things together 16:10:02 one more thing regarding small patches 16:10:15 strigazi: do you have a sense which ordering would be more efficient in terms of work? 16:11:07 my gut sense was that flatten attributes first would advance us sooner toward the vision 16:11:10 I was thinking to break the patch in pieces and instead of adding new db migrations iterating on the latest one to reach the desired schema 16:11:45 that sounds okay to me 16:11:54 others? 16:12:03 Which migrations? 16:13:13 Eg I could drop the _id suffix in a later patch and first move the attributes with the current names to cluster 16:13:17 then 16:13:35 in the same migration reduce the number of attributes to a dict 16:14:09 strigazi: can you explain why you want the migration steps to be compressed in that way? 16:14:10 basically ClusterAttributes is our dict 16:14:33 Are you saying blobify our attributes now? 16:14:56 As a second step 16:15:47 adrian_otto to avoid schema transformations since it's one feature. This was the initial plan. 16:16:05 I haven't investigated if transformations affect performance 16:16:09 This might make migrating attributes to nodegroups harder 16:16:27 Drago the blob? 16:16:38 or migrations? 16:16:38 yes 16:16:54 ok 16:17:14 But maybe not much harder, just something to consider 16:17:45 If attributes are in a blob in the db, you'll have to write something that moves them out into the nodegroup attribute blob 16:18:56 OK, seems that I'll lose time on that. I'll move with the proposed approach and move all attributes to cluster_attributes 16:19:32 any more remarks on the Flatten Attribute blueprint today? 16:19:41 We should concentrate on the nodegroup migrations as early as we can 16:20:18 o/ 16:20:23 Drago: before Driver + config? 16:20:30 juggler: hi 16:20:51 That's a good question, because they'll affect each other 16:20:55 adrian_otto: hello 16:21:01 I'll have to think on it more, some other time 16:21:10 Drago: right. 16:21:13 #link https://blueprints.launchpad.net/magnum/+spec/nodegroups Nodegroups [Drago] 16:21:28 No update from me, I've just been chasing Kubernetes bugs... 16:21:42 #link https://blueprints.launchpad.net/magnum/+spec/cluster-upgrades Cluster Upgrades [strigazi] 16:21:53 Heh, anything from you jvgrant_? 16:21:53 thanks Drago 16:22:34 https://review.openstack.org/#/c/439849/ 16:22:42 not on nodegroups, created a WIP spec for driver resource though that will be needed by nodegroups and cluster upgrades 16:22:49 https://review.openstack.org/#/c/439849/ 16:23:33 Nice! 16:23:49 yes, indeed. Nice start. 16:24:13 wtg jvgrant_ 16:24:28 should have an update today or tomorrow, then is should be ready to remove WIP 16:25:05 good progress. 16:25:26 Any more questions, remarks, or discussion on the subject of Cluster Upgrades? 16:26:10 Other Work Items 16:26:45 (attendees may identify any work items needing team discussion now) 16:27:12 PTG recap on the ML 16:27:20 Heat did one 16:27:36 Yes, I can take an action item for that 16:28:07 #action adrian_otto to publish a PTG summary to our ML 16:28:18 Thanks! 16:28:22 my pleasure 16:28:33 Would Would someone be interested in debugging the multinode job? 16:29:04 strigazi, i can look 16:29:07 We need some changes in how we run the tests. 16:29:09 did it fail? 16:29:24 The setup was complete 16:29:42 but we might need to change how we trigger the tests 16:29:56 or some other config 16:31:17 ok thanks yatinkarel for stepping up 16:31:27 #topic Open Discussion 16:31:36 I have two notes from previous meetings 16:31:50 one is about changing to cycle-with-milestones 16:32:14 nobody has raised any concerns about that change 16:32:32 so I'm planning to submit the patches to change it 16:32:46 see me if you have concerns 16:33:07 the second is about the architecture diagram we have on our wiki page 16:33:24 danil, one of our contributors worked on an updated version 16:33:46 but I don't think we have the source document that generated the image in order to pick up and finish that work 16:33:59 so we might need to start over 16:34:01 thgouths? 16:34:07 thoughts? 16:34:24 I have a powerpoint slide of the diagram that I did for the OSA workshop 16:34:49 If I remember correctly, it's pretty close to what danil had and you and randall gave feedback on it 16:34:52 is there a link somewhere to the image? 16:35:09 juggler: http://imgur.com/a/af5BP 16:35:13 that's the proposed revision 16:35:27 we have another problem as well 16:35:37 the wiki will not allow for any new images to be uploaded at all 16:35:54 Well that's dumb 16:35:55 so if we do put up a new image, it can't be in-line, it would need to be a link to a diagram 16:36:06 I have a not so attractive diagram in slide number 3 https://docs.google.com/presentation/d/18v8iJkamBGfSamuaVTm8BqGAQGXtf7fV7PxPbCxKcJs/edit#slide=id.g20c51cee01_0_409 16:36:07 that's peculiar 16:36:09 yeah, that's a battle we need to take up with the wiki maintainers 16:36:29 I think we should talk to them, because that's just silly 16:36:35 file uploads were disabled in response to abuse 16:37:04 but we should come up with some way to whitelist users who are allowed to. 16:37:54 any idea what app the original used to make the diagram? Dia? other? 16:38:05 or temporarily enable the feature while we make changes, and disable it again when we are done 16:38:16 original developer 16:38:18 Here's mine http://imgur.com/AxQ8uPY 16:38:42 We can update the colors or whatever. I can email you adrian_otto 16:39:01 Drago: I like that one 16:39:04 tx 16:39:10 yeah the sharp red vs blue is a little harsh to me 16:39:22 a little :) 16:39:39 red aside, it's a great diagram 16:39:43 ^ 16:39:54 Magnum too much for you?? 16:40:03 hehe 16:40:22 red=stop 16:40:23 lol 16:41:03 adrian_otto: Sent! 16:41:11 I think we should add the native clients somehow 16:41:16 tx Drago 16:41:27 strigazi: good suggestion 16:41:32 thx Drago 16:41:32 adrian_otto: If you find updating it to be a pain, I can do it 16:41:41 + DC/OS + Cluster Driver 16:41:43 I know how to use powerpoint ;-) 16:41:48 Drago, little update to base OS supported currently, rkt/docker etc 16:42:06 + baremetal 16:42:12 ubuntu meos is missing 16:42:18 what's powerpoint? :) 16:43:18 juggler: pay no attention to the man behind the curtain 16:43:19 yatinkarel: Oh, sure. I mostly fixed the current arch image, changing it as little as possible. It should be updated with more generic things 16:44:05 Drago, +1 yes we can incorporate at least what magnum supports currently 16:44:40 and update as magnum advances 16:45:01 Drago: received. 16:45:31 I have a patch in review for last few weeks. could you guys take a look and let me know your comments https://review.openstack.org/#/c/427507/ 16:45:57 That poor patch... 16:46:21 What tests are flaking? 16:48:13 Drago: without this patch admin operations wouldn't work 16:48:30 Drago: or at least would not work correctly 16:50:56 Right 16:51:03 ok, should we end a little early today? 16:51:32 Did Drago get an answer to his earlier q? 16:51:54 I can't remember the question so it must not have been that important 16:52:05 Drago> What tests are flaking? 16:52:09 Unless it was about flaking tests. I'll just ask vijendar directly 16:52:27 adrian_otto: I think so 16:52:43 ok, pausing for just more more moment... 16:53:16 Thanks for attending everyone. Our next team meeting will be at 1600 UTC on 2017-03-14 in #openstack-meeting-alt 16:53:18 ο/ 16:53:21 #endmeeting