16:00:03 #startmeeting kolla 16:00:08 Meeting started Wed Mar 21 16:00:03 2018 UTC and is due to finish in 60 minutes. The chair is Jeffrey4l. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:11 The meeting name has been set to 'kolla' 16:00:19 #topic Roll-call 16:00:23 o/ 16:00:23 o/ 16:00:27 o、 16:00:39 0/ 16:01:05 o/ 16:01:08 o/ 16:01:15 o/ 16:01:23 o/ 16:01:30 o/ 16:01:46 #topic announcements 16:02:08 first of all, welcome caoyuan to join the core team 16:02:08 o/ 16:02:16 :) 16:02:21 thanks 16:02:29 congratulation caoyuan 16:02:31 grats caoyuan 16:02:41 congrats ;) 16:02:53 thanks :) guys 16:03:13 o/ 16:03:19 second, kolla rc2 is tagged. and kolla 6.0.0 tag will be released soon. 16:03:32 any announcements from community? 16:04:16 #topic ptg summary bp 16:04:23 #link https://etherpad.openstack.org/p/kolla-rocky-ptg-recap 16:04:37 thanks pbourke for preparing the summary. 16:04:54 we need convert them to bp. 16:04:58 please open the link. 16:05:35 and add proper bp for them, if the pb already exists, please just add the link. 16:05:47 a handful have already started bps 16:05:50 will add links now 16:06:06 we will be after 10 min 16:06:52 if anyone is interested in any bp, please assign it to yourself. 16:07:30 I am working on 3rd party vendor plugins for Neutron 16:07:52 cool. 16:07:59 that approach could be used for plugins that are currently in tree and are planning to be split out 16:08:17 and I understand the issue that Sam Betts brought up during the PTG 16:09:01 does a BP already exist for this? 16:09:02 sadasu, could you write a spec or POC to explain what the issue and how to solve it? 16:09:38 Jeffrey41: yes, if non already exists, I would be happy to start a new spec 16:10:00 iirc, there is no such a bp 16:10:40 I also have some initial changes out in https://review.openstack.org/#/c/552119/ 16:11:12 let me first create a spec and then link this review to that 16:12:30 thanks to caoyuan for providing comments on the review 16:12:49 sadasu you are welcome:) 16:15:50 regarding backups, i've added the link in to the existing bp for mariadb backup and recovery 16:16:00 yankcrime, thanks. 16:16:29 time is up. thanks all guys. 16:16:45 please feel free to take the pbs. 16:17:08 and if you have any concern, please throw it out in meeting or ML. 16:17:35 anythink wanna be talked for the pb now? 16:18:11 guess no 16:18:14 let us move on 16:18:15 pb? 16:18:21 yes. 16:18:32 what is that? 16:19:09 any bluesprint in the etherpad you wannt to talk here, now? 16:19:43 Jeffrey4l: you mean bp (not pb)? 16:20:04 hrm, sorry, my bad 16:20:08 ah right 16:20:11 one thing 16:20:24 sometimes there's confusion on who can set those to approved 16:20:30 could you clarify Jeffrey4l ? 16:21:10 Jeffrey4l: I would be working on rolling upgrade for few of core Services and I would assign the same 16:21:24 I think its anyone in the kolla launchpad group, but Im not sure what makes a bp eligble for approval 16:21:26 normally, another core member, who is not the drafer, could. 16:21:34 ok 16:21:49 we are not that strict on this. 16:21:57 good enough for me :) 16:22:10 in other team, a spec and a roll-voting is required. 16:22:23 and we only do this for big change. 16:22:34 spsurya__, cool, thanks. 16:22:52 I can see a lot of bp have old series version. Maybe try to clean bp ? 16:23:12 ktibi, yep. we need update them. 16:23:27 spsurya__, thanks, I'll update the current status, I also working on Heat locally 16:23:40 spsurya__ cool and the other's rolling upgrade I glad to complete 16:24:08 duonghq: caoyuan_ roger 16:25:07 feel free to re-target the bp status and series. 16:25:12 ok. let us move on 16:25:18 #topic kolla-kubernetes project split 16:25:19 caoyuan_, spsurya__ Glance ps is finished, can you test with me? 16:25:45 this is still WIP 16:25:47 #link https://review.openstack.org/552531 16:26:04 i think we guys are agree on that split the kk8s project. 16:26:16 duonghq my pleasure 16:26:22 are any kolla-k8s cores here? 16:26:24 duonghq: sure, we can discuss in opendiscussion 16:26:28 portdirect, anything you wanna to update now? 16:26:46 portdirect, tbh, the former active contributor are not here. 16:26:56 i just read portdirect's comment and I think it makes sense 16:27:39 :( 16:27:54 i checked the code of kolla-k8s and osh ( shamed on me that i never tried them), i am agree with the former kk8s active contributor, 16:28:02 portdirect: usually agreement is a good thing ;) 16:28:23 the microservice is a good arch than only one charts. 16:28:36 As someone who used to be active in the project I'm just worried that it would be confusing, and potentially damaging, to 'launch' a project that from the outset does not have a path towards sustainability. 16:29:22 portdirect, it is hard to upgrade the openstack through only one charts for nova. 16:29:34 I contest that :) 16:29:34 portdirect, spsurya__ and i will start contribute this project 16:29:59 and we will also try to find other volunteers 16:30:04 Jeffrey4l: IIRC rwellum is also in 16:30:14 also osh supports the same architecture as kolla-k8s if desired: https://github.com/openstack/openstack-helm/blob/master/nova/values.yaml#L1594-L1639 16:30:17 yeap 16:30:18 why not just contribute to osh 16:31:03 Jeffrey4l: duonghq has also shown interest at a time don't know at the moment 16:31:08 I dont want to cause to much issues - but its worth noting that osh has three large orgs in or close to prod 16:31:11 hrm, i see, interesting, 16:31:34 and for them, upgrade was the primary attraction 16:31:43 pbourke, it is mainly concern with arch or deploy philosophy 16:32:15 I imagine those issues could be worked out with osh 16:32:25 assuming they even are issues from an end user perspective 16:32:26 Jeffrey4l: would it help if i produced a demo of OSH run as micro-charts? 16:32:51 i think i get your idea. 16:33:06 use variables to control which chart is loaded in heml 16:33:08 helm* 16:33:31 yup 16:33:32 i missed that when i review the code. 16:33:52 this provides the exact same flexibility as the kolla-k8s deployment method 16:34:06 where each object can be addressed individually 16:34:39 there is another thing i concern. 16:35:05 how helm load the *.yml? any possible order? 16:35:22 as in the templates? 16:35:50 yes. 16:36:32 it uses a semi-deterministic order 16:36:43 which is controlled through one of two methods 16:37:09 okay, i will check that 16:37:12 we use dependency/environment checking extensivly: https://github.com/openstack/openstack-helm/blob/master/nova/values.yaml#L205-L219 16:38:15 roger thanks 16:38:18 or you can take control of the manifest deployment via toggling them in the over-rides in a playbook 16:38:33 understand. 16:38:33 *via a playbook 16:38:59 at least for microservice, osh and kk8s are almost the them. 16:39:40 mybe i need dig these two projects more later. 16:39:58 we choose a different `default path` but you can drive them in veritually the same way if desired 16:40:18 its also worth noting that with osh - we fully support both Kolla and LOCI images 16:41:10 taking the same path that Kolla-Ansible is now exploring for bind mounting in the config and scripts 16:41:58 got. 16:43:17 thanks 16:43:24 :) 16:43:26 i am agree now that we should and may be co-operation. but we still need some time before make the decision. ;D 16:43:32 ++ 16:44:05 thanks portdirect 16:44:24 anything else about kk8s and osh? 16:44:52 let us move on 16:44:55 #topic Documentation RST convention upgrade 16:45:01 chason, your floor 16:46:06 seem he is not here. 16:46:08 #link https://review.openstack.org/#/q/status:open+project:openstack/kolla-ansible+branch:master+topic:bp/optimize-the-documentation-format 16:46:22 this is a legecy patch in fact. 16:46:43 there some syntax or convention issue in doc 16:46:54 https://review.openstack.org/#/c/538220/ 16:46:55 feel free to review them 16:46:56 thanks 16:47:37 #topic Open Discussion 16:47:43 any volunteers? 16:48:08 if we have time i wouldn't mind getting any feedback on my email to the mailing list regarding upstreaming the kollacli 16:48:16 and putting it under kolla management 16:49:03 im good with it but a little biased obviously ;) 16:49:04 we can create a new repo for it. 16:49:41 do we need a spec? 16:49:51 duonghq: i can help in test and also would like add script for nova rolling upgrade 16:50:00 would like people to be in agreement with the overall way it works before importing it 16:50:32 well, since it is already sort of fully implemented and i linked the docs in the email to the mailing list, i'm not sure what extra a spec would provide. 16:50:37 would be nice to see a spec for the cli, but clearly an implementation exists so how much will you be willing to change? 16:50:50 spsurya__, nice, thank you very much 16:50:57 i am fine with it changing in whatever ways the community wants, that is the whole point of doing the upstream :) 16:51:10 mgoddard ^^ 16:51:19 another approach could be - import existing as a pre 1.0 release, with the proviso that anything may change before 1.0 16:51:23 hope that we can finished 1st version for rolling upgrade of all core services, it will need extensive test 16:51:43 mgoddard: seems a good idea 16:51:44 how about move the project into openstack project namespace, then if there is anything not properly, we can improve it through a spec 16:52:15 mgoddard: that is fine with me as well... consider the cli pre-release until a point that the community is happy with it. 16:52:33 duonghq: will learn in advance way to test for glance probably tomorrow 16:52:44 spsurya__, nice 16:52:58 on the question of a new repo vs. in kolla-ansible - how closely tied is it to the inventory, config, playbooks etc.? 16:53:11 bmace: people may be more likely to read a spec rather than trawl the oracle docs 16:53:26 we also need to optimize the solution of zedro downtime of ks 16:53:27 prob not a big deal though 16:53:38 most seem in favor so 16:53:48 mgoddard: internally we have a totally separate repository for the cli, so having it broken out is fine, but for the cli to be actually usable, it does rely on bits from kolla-ansible, like the passwords, playbooks, inventory, etc. 16:53:50 pbourke, the issue is we can not wait until the spec is merged. 16:53:56 if the cli is in a separate repo, then does the playbook 'interface' need to become stable? 16:55:25 for ansible playbooks, the interface is more like "inventory" file or "globals.yml" 16:55:29 mgoddard: a problem with the separate repo is that conceptually a change in kolla-ansible could break the cli, though in our experience it would need to be a pretty big change and that issue happens very rarely 16:55:52 well I guess we'll have tests and gates to prevent that too 16:55:53 bmace, we can add cross-project jobs to ensure this. 16:56:02 mgoddard: i think in the long run that could be covered by the introduction of some sort of co-dependent tests. 16:56:14 Jeffrey4l: exactly my thought, right 16:56:22 after the kolla-cli, we can use it in kolla-ansible jobs to genreate the inventory or global.syml file.s 16:56:51 it probably shouldn't come up often, but it just means we can't do something like rename site.yml to somethingelse.yml 16:57:12 all that can be configurable 16:57:44 mgoddard, once that done, jobs failed and we could fix kolla-cli at the same time. 16:57:45 mgoddard: right, some sort of thought of the effect on the cli needs to be kept in mind, which i think is good all around, because some people use it already without a cli and might do scripts of their own and similar changes could break those people also. 16:58:10 that's true 16:58:18 let us start by separate repo. once it is not proper in the future, we can still merge two repo into one 16:58:49 this is not a big deal. 16:58:51 Jeffrey4l: that sounds great. starting with it broken out, as it is internally for us, should also make the transition easier. 16:59:18 yeah. 16:59:42 ok time is up. let us move to kolla channel 16:59:49 is there any way that i can be a core for just the cli? i'm not sure how it was for the kolla-k8s stuff, i think something was done similar? if not that is fine.. i just know i have a bit of outstanding work to do on it and will just need to get other folks to review it :) 16:59:49 ok 16:59:57 thanks guys for coming. 16:59:57 kk 17:00:06 thanks 17:00:08 #endmeeting