16:00:45 #startmeeting kolla 16:00:46 Meeting started Wed Jan 17 16:00:45 2018 UTC and is due to finish in 60 minutes. The chair is Jeffrey4l. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:48 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:50 The meeting name has been set to 'kolla' 16:01:01 #topic rollcall 16:01:07 o/ 16:01:12 o/ 16:01:14 o/ 16:01:30 o/ 16:01:30 o/ 16:02:10 o/ 16:02:17 #topic announcements 16:03:01 any annoucements from community? 16:03:14 o/ 16:03:28 guess no 16:03:33 lets move on 16:03:39 o/ 16:03:41 #topic Alpine Linux base distro (chmarkus) 16:03:47 chmarkus, yours 16:04:01 ok 16:04:35 So far, I was able to provide basic Alpine support for both base and openstack-base 16:04:56 that excludes the 3rd party repos from openstack-base (ceph, galera, etc.) 16:05:27 ceph and galera is necessary 16:05:40 I'm aware of that 16:06:17 I tested it by building a barbican service ontop of the alpine image using kolla and put it into an existing OpenStack infrastructure which uses ubuntu based kolla images otherwise - this worked fine for both manual and tempest tests so far 16:06:30 so simple API services and the like are possible 16:06:48 however the showstoppers here are dependencies like ceph 16:06:54 which is needed for compute 16:07:14 ceph is not available for alpine and from quick research there still seem to be conflicts 16:07:48 another thing is libguestfs - which is requested on the alpine tracker but has not been packaged yet - it is presumably necessary for KVM support of compute iirc 16:08:15 chmarkus, libguestfs is only need when using "libvirt_inject_xx" 16:08:24 and this feature is removing 16:08:28 okay 16:08:51 one showstopper less I guess 16:08:58 but still ceph is a problem 16:09:26 yes. lots of openstack env depends on ceph. 16:09:47 also from my experience, Qpid support (rabbitmq alternative) is not possible due to proton-c not compiling (during the pip install process) on alpine due to musl libc problem 16:10:08 which was part of the openstack-base template as well, so I had to omit that also 16:10:26 so in conclusion, alpine is not there yet unfortunately :/ 16:10:41 nice jobs chmarkus 16:10:43 not sure which services uses that library, we do not have qpid support 16:10:57 the plan to provide a proof-of-concept with the compute services sadly did not come to realization 16:11:28 egonzalez, check https://review.openstack.org/#/c/494519/ 16:11:59 do you notice the size of the built alpine images? 16:12:16 is it small as expected? 16:12:55 sec 16:13:25 the problem is that openstack-base preloads everything that might potentially be needed by any service ontop of it - so it is still big - 576MB for alpine-source-openstack-base 16:14:04 yes. but this reduce the total size of all images. 16:14:16 it does yea 16:14:44 however, this would require all images to be compatible with the alpine base 16:15:03 which in the current state would not be possible from my pov 16:15:27 anyway, thanks chmarkus for the testing. 16:15:41 and kolla won't support alpine image. 16:15:55 would it be of any use to provide the code for alpine+source for base and openstack-base upstream to you? 16:16:08 i guess not huh? 16:16:23 upstream? what that mean? 16:16:48 compile ceph from code? 16:16:52 the code adjustments I did for the alpine support for those images 16:17:05 in my clone of the kolla repo 16:17:16 i guess no. no one will use that. 16:17:31 what about the blueprint for alpine? 16:17:39 should that be updated/rejected? 16:17:45 i think we can close it with what you found. 16:17:59 what do you think? chmarkus 16:18:22 Sadly, that's the only conclusion I can draw from my experience yes. 16:18:51 nice work chmarkus 16:18:54 cool. i will write some comments and close it later. 16:18:57 thanks chmarkus 16:19:05 should I put some information there? 16:19:09 good work indeed 16:19:10 anything else chmarkus ? 16:19:14 chmarkus: you should 16:19:16 chmarkus, sure. 16:19:17 put your conclusions 16:19:31 chmarkus: makes Jeffrey4l's job easier 16:19:33 :) 16:19:33 ok how do I edit it? ^^' 16:19:38 yep ;D 16:19:41 thanks for your work chmarkus :) 16:20:00 chmarkus: you can edit the whiteboard , there is an edit button 16:20:02 if you log in 16:20:03 chmarkus, let's talk this later. i am not sure whether you can do this now. 16:20:12 let's move on? 16:20:15 sure 16:20:15 ok 16:20:20 thanks. 16:20:25 thank's for hearing me out :D 16:20:37 #topic open discussion 16:20:44 any topics? 16:20:56 o/ 16:21:03 duonghq, please 16:21:26 thank Jeffrey4l 16:21:46 first thing I want to bring to this meeting is work from my colleague (daidv) 16:21:47 https://review.openstack.org/#/c/531641/ 16:21:56 #link https://review.openstack.org/#/c/531641/ 16:22:16 it's 2nd of policy.yaml supporting series 16:22:27 the 1st one is merged 16:22:36 o/ when duonghq is done 16:22:38 ( the # line is not necessary now :) ) 16:23:07 yes. this is nice and we should merge it. 16:23:10 so please review it due to it's one of Q goal for OpenStack services 16:23:23 *many OpenStack service indeed 16:23:26 the patches are big but easy to review. 16:23:51 wait, found a typo 16:24:01 yep, but part 8 is different from other 16:24:32 #link https://review.openstack.org/#/q/topic:bp/support-custom-policy-yaml+(status:open+OR+status:merged) 16:24:43 feel free to review this ^^ 16:25:13 Jeffrey4l, thanks 16:25:19 i see, part 8 copy all policy file to horizon 16:26:01 Jeffrey4l, it's quite painful to make Ansible happy 16:26:33 but you did it ;D 16:26:51 my colleague and me did it :P 16:26:59 cool. thanks duonghq 16:27:04 anything? 16:27:24 the second topic I want to take is about rolling upgrade 16:27:38 egonzalez, thank for your review 16:27:51 so, can I go to my 2nd topic? 16:28:54 yes. please duonghq 16:29:27 I have some ps on the review queue, the oldest is neutron one: 16:29:31 #link https://review.openstack.org/#/c/407922/ 16:29:40 rolling upgrade is very hard. i still have on courage to facing it ;D 16:29:53 but it's a bit hard to debug so I move to next project: cinder 16:30:16 #linke https://review.openstack.org/#/c/529546/ 16:30:19 #link https://review.openstack.org/#/c/529546/ 16:30:46 I see the ansible(-image) is merged, thank egonzalez and sp for it 16:31:04 so I hope that somebody can review the kolla-ansible part 16:31:11 (ansible-image -> kolla-image) 16:31:46 duonghq, re neutron upgrade, how works with plugins? 16:32:00 are they upgraded after server or is part of all db sync 16:32:02 added to my review list~ 16:32:05 I also pushed the glance ps but I face bug from glance size, so I think I'll set it aside for awhile 16:32:15 *size -> side 16:32:18 Jeffrey4l, thanks 16:33:05 I mean, wont be any conflict when having n-server db in queens and networking-sfc/fwaas/vpnaas in pike? 16:33:32 egonzalez, it should be upgrade after n-server is upgraded as neutron doc points out 16:33:34 Once neutron-server services are restarted with the new database schema and the new code, it’s time to upgrade Neutron agents. 16:33:36 https://docs.openstack.org/neutron/latest/contributor/internals/upgrade.html 16:33:58 the n-server upgrade contains database migration 16:34:08 egonzalez, it depends on the implementation on the plugins side. if it implement version objects, then it should be OK. 16:34:18 if not, we can do nothing for this ;( 16:35:21 duonghq, will test the networking-sfc upgrade case at let you know 16:35:42 Jeffrey4l, egonzalez, thanks very much 16:35:51 that's all from me 16:36:07 o/ 16:36:22 thanks duonghq , please rwellum 16:36:57 i check the networking-sfc's code. version object is not implemented. i guess upgrade n-server will break networking-sfc. 16:37:06 Can we have some cores take a look at: https://etherpad.openstack.org/p/kolla-rocky-ptg-planning - and add achievements and goals for kolla/kolla-ansible? Otherwise there'll be nothing to discuss :) 16:37:27 0/ 16:37:53 sorry guys a bit late 16:38:15 hi spsurya_ 16:38:18 rwellum, thanks. 16:38:37 thanks Rich 16:38:41 yes. this page should be filled before the ptg. 16:38:45 Sure that was it from me 16:38:51 hey spsurya_ D 16:39:10 should we do a brainstorming and then select which topics discuss at PTG? 16:39:44 rwellum, we will add this into the next weekly meeting. 16:39:53 thank Jeffrey4l for information, it's a bit more painful here if it's in this case :P 16:39:58 egonzalez, +1 16:40:14 egonzalez, +1 16:40:28 since we only have 20 min. if there is no other topic, 16:40:48 egonzalez: +1 16:41:07 let's focus on the Rocky goals for "kolla" repo projects. 16:41:29 what the painful and what you want to added to kolla? 16:41:34 feel free to add it . 16:41:39 or talk here. 16:41:43 +1 16:44:25 For kolla-k8s - it's a little hard to discuss without kfox/jascott around as they have a few things they are working on/planning. 16:44:46 So can defer that until I catch up with them. 16:44:52 the ever recurring topic "plugin split from main images" 16:44:59 rwellum, sure. 16:45:27 egonzalez, what's this topic meaning? 16:45:54 I think the result is our images are much smaller 16:45:59 how to handle plugin installation in images (ie neutron ovn) 16:45:59 without unused stuff 16:46:28 duonghq, the current result or the expectation? 16:46:51 Jeffrey4l, then expectation 16:47:07 #link https://etherpad.openstack.org/p/kolla-queens-ptg-images 16:47:08 if we can factor out the "unused" plugin 16:47:56 in fact the "unused" plugin do not use much size. 16:48:10 just several python text file. 16:52:22 yup 16:56:17 ok, we will focus on the page next meeting. 16:56:32 nice 16:56:43 will you go to the PTG Jeffrey4l ? 16:56:51 duonghq, no :( 16:56:58 will you ? 16:57:23 Jeffrey4l, I'm waiting for the TSP result 16:57:45 cool hope you can got it. 16:57:52 thanks 16:57:55 anyway, we are running out of time. 16:58:14 thanks for everybody coming. let end the meeting 16:58:17 thank you guys 16:58:23 thanks 16:58:23 and have a nice day/night 16:58:29 #endmeeting