16:03:21 #startmeeting openstack_ansible_meeting 16:03:21 Meeting started Tue Apr 2 16:03:21 2019 UTC and is due to finish in 60 minutes. The chair is mnaser. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:03:22 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:03:23 o/ 16:03:25 The meeting name has been set to 'openstack_ansible_meeting' 16:03:32 o/ 16:03:58 hi all 16:04:12 o/ - means smth like "I'm here"? :) 16:04:13 https://etherpad.openstack.org/p/osa-train-ptg 16:04:15 reminder as per usual ;) 16:04:35 \o/ 16:06:02 o/ 16:06:31 o/ 16:10:42 o/ 16:11:12 #topic office hours 16:11:25 I mean, yay, we have stein, it's production ready, I trust it 16:11:32 our sjc1 region is running it, openstack CI is using it now :) 16:12:41 so I think we should be able to confidently release on release day 16:15:13 that's awesome mnaser ! 16:15:16 thanks for everyone involved 16:15:22 nothing better then testing in production 16:16:33 other than that does anyone have any particular things they want to discuss? 16:17:04 guilhermesp: :D 16:17:08 it's called CI/CD 16:17:13 :D 16:17:25 mnaser: yes 16:17:41 I think we should, if possible, timebox the ptg: https://review.openstack.org/#/c/648947/ 16:17:43 woops 16:17:50 https://etherpad.openstack.org/p/osa-train-ptg 16:18:11 but there is no urgent need for it, just saying 16:18:24 yes that would be good for us who have to jump around 16:18:38 next to that, the list of bugs grows. I didn't generate it this week, but it's massive 16:18:52 i think now we have just some role to retire too 16:18:54 so I think it would be nice to cleanup the bugs before release 16:19:01 but I think it's not too urgent 16:19:12 yeah and that take a few cycles 16:19:13 :P 16:19:39 yeah hmm 16:19:47 wanna build a list anyways so we can see what we're working with? 16:20:03 guilhermesp: ^ ? 16:20:17 the roles to retire are here https://etherpad.openstack.org/p/osa-python-build-simplification 16:20:26 but we can make a different etherpad tho 16:21:16 I think non integrated roles should no longer exist with our new CI infra 16:21:24 its either integrated and tested with integrated or its not 16:21:32 I agree 16:21:48 afaik monasca* are not integrated yet 16:22:23 we also have a leftover of the python_venv simplification https://review.openstack.org/#/c/648583/ 16:23:07 actually, this is use-integrated-repo-for-all-tests topic 16:23:08 yeah I had no idea what to do with trove at that point 16:23:40 yes, it is requiring some overrides in role tests, and as we are doing integrated tests 16:23:51 I also have no idea how to define those overrides in integrated repo 16:25:41 guilhermesp: we can use the user_variables.aio.j2 or something like that 16:25:44 its in the bootstrap host role 16:26:18 huum good call mnaser so I think worth it to take care of this before we release? 16:27:11 trove required trove_service_net_setup to true but it's false by default, is that expected to be in defaults? 16:27:27 * guilhermesp asks to trove maintainers :) 16:28:25 yeah im not sure, maybe whoever added that role can bring it up 16:28:50 or I can push up a patch and set it to true and then everyone is happy 16:30:33 guilhermesp: if its only for aio it probably doesn't hurt 16:31:25 if is only for aio I think just defining to true in user.variables.aio.j2 would be good 16:32:36 let me patch it here and add a depends-on 16:35:01 ok great 16:37:28 other that that I think its the quiet after the storm :) 16:41:15 Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible master: Add trove_service_net_setup to aio deploy https://review.openstack.org/649397 16:41:17 mnaser: ^ 16:41:40 guilhermesp: good stuff 16:42:14 im pretty happy, we'll have a good wind-down after rc1 16:42:14 now depends-on here and let's see https://review.openstack.org/#/c/648583/ 16:42:38 Mohammed Naser proposed openstack/openstack-ansible-os_trove master: Test using integrated build https://review.openstack.org/648583 16:42:46 maybe we can do rc2 if we land the trove stuff 16:43:18 yeah, as is the last thing related integrated tests I believe 16:44:13 mnaser: I think you added the wrong link in the depends-on 16:44:28 ah yikes 16:44:29 you're right 16:44:51 Mohammed Naser proposed openstack/openstack-ansible-os_trove master: Test using integrated build https://review.openstack.org/648583 16:45:23 nice, let's sit and watch 16:46:36 other than that I'm very happy and satisfied of our release. I think the next one will be ever better with some of the role clean up ideas I have 16:47:09 I will do my best to be even better 16:47:34 I think decreasing the # of variables 16:47:39 to make it easier to configure 16:47:49 get rid of any `shell` references 16:48:08 Probably we may improve work with hosts? 16:48:28 and proper ordering: deploy venv, configure db and sync, start services, add in keystone 16:48:39 sometimes the order is wrong and then the logs are spammed with like "Service down" 16:48:41 or something like that 16:48:50 noonedeadpunk: what do you mean about that? 16:49:13 Like moving from generating it with bash script to generating with ansible template 16:49:19 or smth like that 16:50:11 yeah 16:51:29 oh you mean the /etc/hosts file 16:51:30 ah yesss 16:51:36 Yep:) 16:51:45 I think the idea of generating it was because it was some performance thing 16:51:45 so those ideas already in etherpad mnaser ? 16:51:52 I put some there yes 16:51:55 noonedeadpunk: maybe you wanna add that one 16:51:56 nice 16:52:02 https://etherpad.openstack.org/p/osa-train-ptg 16:52:45 so I believe the some of the role retirement are going to be done next cycle... but not sure if is too important to add as a discussion item 16:52:50 in etherpad 16:53:21 mnaser: if it was some previos decision I'm not really aware of (especially in terms of performance), than there's probably no reason for that? 16:53:35 noonedeadpunk: well a lot of the stuff in OSA was to optimize for older releases of ansible 16:53:40 ansible has come a long way and it has a lot of new things now 16:53:52 like for example, action_modules .. we can use one to build the /etc/hosts locally, and then ship it out 16:54:06 Yeah, for older realeases it even might be even not really possible/// 16:54:07 I think logan- knows some of these things 16:54:22 i think we should have a really good bash at getting mitogen to work 16:54:44 jrosser: yes, I added that too 16:54:45 mitogen ++ 16:54:53 becasue with the best will in the world the amount of work that needs to be done in the tasks is finite and can't be reduced much 16:55:06 but the ansible overhead is another matter 16:55:08 its in the ideas of like eliminating stuff that we're maintaining and deferring to tools that are really good at doing what they do 16:55:28 so why maintain our own ssh connection driver when mitogen can give us what we want for example 16:55:35 in a much more efficient manner too 16:56:00 we need to engage with dw a bit, there are still some things to fix on his side 17:02:36 jrosser: im up for us to start doing that, I think he'll be happy to see an open source consumer 17:02:40 #endmeeting