14:02:47 <tellesnobrega> #startmeeting sahara
14:02:47 <openstack> Meeting started Thu Mar  7 14:02:47 2019 UTC and is due to finish in 60 minutes.  The chair is tellesnobrega. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:50 <openstack> The meeting name has been set to 'sahara'
14:02:53 <tellesnobrega> sorry I'm late
14:03:02 <tosky> o/
14:03:08 <jeremyfreudberg> o/
14:04:17 <tellesnobrega> #topic News/Updates
14:05:40 <jeremyfreudberg> working on dashboard stuff and writing a candidacy statement
14:05:45 <tosky> \o/
14:06:16 <tosky> I mostly focused on improving the set of jobs that we run on the gates, (v2 jobs, cleanup of dashboard jobs, jobs for the plugins, etc)
14:06:20 <tellesnobrega> well written btw
14:06:35 <jeremyfreudberg> thanks! :)
14:06:38 <tellesnobrega> I'm still fighting py2 -> py3 compatibility
14:07:15 <tellesnobrega> I can get in more details later
14:08:38 <tellesnobrega> any more news?
14:09:10 <tellesnobrega> lets move on
14:09:18 <tellesnobrega> #topic APIv2
14:09:29 <tellesnobrega> how are we doing for making apiv2 stable today?
14:09:38 <jeremyfreudberg> let's go for it!
14:09:42 <tosky> most of the "deploy with unversioned endpoints" patches have been merged
14:09:49 <jeremyfreudberg> i don't see any reason not to make it stable
14:09:53 <tosky> the only remaining one is for tripleo-heat-templates
14:10:08 <tosky> but I'm going to start a friendly nagging campaign :)
14:10:52 <tellesnobrega> jeremyfreudberg, can you write the patch to make it stable?
14:11:02 <tosky> tellesnobrega: it's already there :)
14:11:08 <jeremyfreudberg> tellesnobrega, https://review.openstack.org/#/c/582282/
14:11:50 <tellesnobrega> oh, missed it
14:12:14 <tellesnobrega> +2ed
14:12:22 <tellesnobrega> thanks for the hard work on it jeremyfreudberg
14:12:29 <tellesnobrega> and tosky
14:12:38 <jeremyfreudberg> my pleasure :) thanks all
14:13:22 <tellesnobrega> tosky, whenever you feel like we are ready to merge it, you can do the honors
14:13:53 <tellesnobrega> moving on
14:14:03 <tellesnobrega> #topic Python 3
14:14:56 <tellesnobrega> so, I can make sahara work with python 2 and with python 3, but not at the same time
14:15:58 <tellesnobrega> and right now, I'm not sure how to get it to work
14:16:16 <tellesnobrega> the issue is with how we load from pickle on the remote machine
14:16:32 <tellesnobrega> on python3 it needs to read from sys.stdin.buffer
14:16:41 <tellesnobrega> and on python2 it reads from sys.stdin
14:17:09 <jeremyfreudberg> "at the same time" means a mixture? or it just means one codebase which supports both individually?
14:17:21 <tellesnobrega> supports both individually
14:17:26 <jeremyfreudberg> got it
14:18:00 <tellesnobrega> I'm thinking that I may have to check the local python version running and send it as kwargs to the remote machine
14:18:39 <tellesnobrega> and update the script to execute the appropriate function depending on the version
14:19:05 <tellesnobrega> does that seem like a good plan?
14:21:18 <tosky> if it works, yes - but how does it work? If you have python3 on the sahara server and python2 on the instance, the instance can just executed one codebase (python2)
14:22:02 <tellesnobrega> I think that the remote python won't be an issue
14:22:14 <tellesnobrega> I just need to get the right "connection"
14:22:26 <tellesnobrega> I will give it a try and see how it goes
14:22:41 <tosky> the complication comes from this cross-python communication
14:23:05 <tosky> other had the same issue; looking around, I can see for example https://review.gluster.org/#/c/glusterfs/+/21320/
14:24:34 <tellesnobrega> hmm
14:24:39 <tellesnobrega> I will take a look into it
14:25:47 <tellesnobrega> hopefully I will be able to get it to work today
14:25:49 <tellesnobrega> lets see
14:26:17 <tellesnobrega> other than that, do we have any blocking issues for M3?
14:28:06 <tosky> not really super-urgent, but we have two reviews for the plugins
14:28:23 <tellesnobrega> which ones?
14:28:27 <tosky> not python3 related (well, one of them is iirc)
14:29:08 <tosky> https://review.openstack.org/#/c/634799/ should be good to go
14:29:26 <jeremyfreudberg> i'll review that
14:29:44 <tellesnobrega> thanks jeremyfreudberg
14:29:55 <tosky> on https://review.openstack.org/#/c/639812/ we had an open question - the fix itself is fine, it does not introduce regressions, I was also wondering if it scales or if it makes sense to introduce a more general mechanism
14:30:28 <tosky> but at this point it would be for Train or much later, if we don't want to bump the requirements so early
14:31:55 <tellesnobrega> I would say lets keep this for now
14:32:04 <tellesnobrega> and revisit it later
14:32:42 <tellesnobrega> actually, the code as is now, is better than before, because it was simply comparing a string to int
14:32:51 <tellesnobrega> making that comparison completely useless
14:34:16 <jeremyfreudberg> yeah, in python 2 aren't all strings greater than any int?
14:34:59 <tellesnobrega> yes
14:35:08 <tellesnobrega> that is why it was a useless check
14:35:31 <tellesnobrega> I didn't know that it was a string until I changed to python3
14:35:48 <tellesnobrega> than the comparison exploded
14:37:18 <jeremyfreudberg> why not do it as 1. try to convert to int 2. if fail, convert it to -1
14:37:47 <tellesnobrega> that works too
14:38:35 <jeremyfreudberg> avoids the py3 problem with uncomparable types. no regression if we catch the exception for casting to int.
14:38:47 <tellesnobrega> does that makes sense to you tosky ?
14:39:05 <jeremyfreudberg> i might be missing one aspect of it, so, please weigh in, tosky
14:39:27 <tosky> sure, no problem
14:39:55 <jeremyfreudberg> give that a try, then (technically it's a bug so it can be done after M3 if there isn't time)
14:40:41 <tellesnobrega> I can get that in quickly
14:40:50 <jeremyfreudberg> cool
14:41:23 <tellesnobrega> I might have some good news
14:41:35 <tellesnobrega> the check for sys.stdin type worked with python 2
14:41:38 <tellesnobrega> let me see python3
14:42:11 <tellesnobrega> anything else on this?
14:43:12 <tellesnobrega> #topic M3
14:43:34 <tellesnobrega> just want to double check what we need to cut on m3
14:43:55 <jeremyfreudberg> let's please release a new saharaclient and bump the minimum for sahara-dasbhoard
14:44:00 <tellesnobrega> and also ask if jeremyfreudberg wants to do the honors and make the patches
14:44:41 <tellesnobrega> jeremyfreudberg, you are almost certainly the next PTL and getting the versions in is part of your job to come
14:44:51 <tellesnobrega> so start early is easier later on
14:45:22 <tellesnobrega> so, if you want, you can be responsible for m3 and later on for the final release
14:45:38 <jeremyfreudberg> i'm planning to disappear awfully soon today (and won't be back till what feels like a bit late), so if there are late merged patches i won't be around to get the right hash. i can propose something for the client now, just to understand how it works
14:46:04 <jeremyfreudberg> and i'll handle it for the final release
14:46:10 <tellesnobrega> jeremyfreudberg, ok, so I can do it for m3
14:46:13 <tellesnobrega> you do the final release
14:46:19 <jeremyfreudberg> yup
14:46:28 <tellesnobrega> thanks
14:46:39 <tellesnobrega> ok, so new client, bump version on the dashboard
14:46:59 <tellesnobrega> do we need a sahara cut? I'm still getting a hang on the new process
14:47:48 <tosky> I don't remember if a release is needed for m3, but we need for sure an initial release of the plugins
14:48:02 <tosky> otherwise we can't patch openstack-ansible to install them :)
14:48:55 <tellesnobrega> ok, should them be tagged as beta?
14:48:57 <tosky> so maybe releasing also a new pre-release for sahara core is not going to be so complicated
14:49:09 <jeremyfreudberg> https://releases.openstack.org/reference/release_models.html
14:49:12 <tosky> I don't remember which is the correct version number at this point
14:49:26 <tellesnobrega> we are cycle-with-rc iirc, so we don't need m3
14:49:30 <tellesnobrega> just RC
14:49:36 <tellesnobrega> and final release for sahara itself
14:49:41 <tellesnobrega> the plugins I'm cutting as beta
14:50:35 <tellesnobrega> anything else?
14:50:55 <tellesnobrega> plugins, client, bump version of client on sahara-dashboard
14:51:01 <tellesnobrega> and I think that is all
14:51:22 <tellesnobrega> good update on python3
14:51:42 <tellesnobrega> it worked with python3 now
14:51:44 <tosky> oh
14:51:50 <tellesnobrega> I will send a patch
14:51:58 <jeremyfreudberg> not just bump version of client on sahara dashboard, i think also release sahara-dashboard for m3 -- if there's a requirements bump, i think it's a courtesy to do it before feature freeze
14:51:59 <tosky> do you mean the mixed python3+python2?
14:52:41 <tellesnobrega> jeremyfreudberg, got it
14:52:50 <tosky> jeremyfreudberg: but we can't bump the client requirement of the dashboard before releasing the client, so not for m3
14:53:28 <tellesnobrega> tosky, yes, same image, ran python2 and python3 locally and the cluster became active
14:53:30 <jeremyfreudberg> tosky, but the client will be released sooner, since I'm writing that patch separately from telles's other releases
14:55:07 <tosky> jeremyfreudberg: so that depends on the speed that it will take to accept the release patch for saharaclient
14:55:43 <tosky> because if we miss this week, next week it's outside M3, so maybe at that point we will go directly to RC1
14:56:29 <tellesnobrega> tosky, jeremyfreudberg if it merges really soon we can bump, if not we can only deffer to RC1
14:56:39 <jeremyfreudberg> sure
14:57:44 <tellesnobrega> we are running out of time
14:58:07 <jeremyfreudberg> https://review.openstack.org/#/c/641690/ - think i did that right
14:58:47 <tellesnobrega> looks good
14:59:22 <tellesnobrega> I guess we can close for the day
14:59:30 <tellesnobrega> thanks everyone
14:59:46 <tellesnobrega> see you all next week
15:00:04 <jeremyfreudberg> thanks
15:00:07 <tellesnobrega> #endmeeting