14:00:03 <zhaochao> #startmeeting trove
14:00:04 <openstack> Meeting started Wed Jun 20 14:00:03 2018 UTC and is due to finish in 60 minutes.  The chair is zhaochao. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:08 <openstack> The meeting name has been set to 'trove'
14:00:22 <zhaochao> #topic roll call
14:01:59 <gang> hi
14:02:03 <songjian1994> o/
14:02:05 <zhaochao> Courtesy ping fanzhang kumarmn maciejjozefczyk songjian1994 wangyao zhanggang
14:02:13 <zhaochao> gang: songjian1994 hi
14:03:39 <zhaochao> let's wait some more minutes to see if any other ones will show up
14:05:02 <zhaochao> OK, let's start
14:05:06 <zhaochao> #topic Rocky goal updates
14:06:05 <zhaochao> I updated the mox-removal patches in trove-dashboard, hopefully we could get them merged soon
14:07:22 <zhaochao> and the cluster modules support in troveclient OSC is finally merged, thanks for yao and gang
14:07:58 <zhaochao> I also noticed several patches of the python3-first topic
14:08:05 <gang> I saw the update, all has passed the zuul gate, I think we could add +2 to merge them.
14:08:49 <zhaochao> yes, the first patch of this serie in trove repo could be merged
14:09:22 <zhaochao> but we have some problems in trove and troveclient for python3 support, which were not known by us before
14:09:50 <gang> zhaochao: you mean the gettext problem?
14:10:21 <zhaochao> gang: no, https://review.openstack.org/#/q/status:open+project:openstack/trove+branch:master+topic:python3-first
14:11:01 <zhaochao> though the checking jobs failed by pylint, the root causes are in trove api and troveclient
14:12:23 <zhaochao> we have some python3 issues inside trove api and troveclient, mostly by the distinction bwteen 'str' and 'bytes' in python 3
14:13:40 <zhaochao> there may be some more furture problems for python3 support, we have to do some investigation on that
14:14:37 <zhaochao> if all these pathes finally get merged, as tox environment defaults to python3, then we could say trove is python3 compatible
14:15:22 <gang> it seems should treat python3 support as the first priority.
14:17:11 <zhaochao> I also think python3 support is important, and I'll try to help to get these patches to be merged, some other patches may have to be submitted and merged before that
14:18:45 <songjian1994> I remember that the community has a time plan for all project support py3?
14:19:12 <zhaochao> as we decided to limit our rocky goals, we may have time to get all these goals finished before rocky release
14:19:49 <songjian1994> ok,I see
14:20:30 <zhaochao> songjian1994: I didn't get a clear plan for that, but it seems py3 support is one of the global goals in Pike(?) cycle
14:21:25 <songjian1994> I remember it was once there, but it is not clear enough
14:22:33 <zhaochao> yes, anyway, it's better the community finnally start this work, and we also found there are problems for py3 support in trove
14:23:24 <zhaochao> let me mark the link about python3 first
14:23:28 <zhaochao> #link https://etherpad.openstack.org/p/python3-first
14:25:17 <zhaochao> that's all from me today. anything updated from you?
14:25:42 <gang> zhaochao: you can list the problems about lack of py3 support in trove, may we could do some work for some of them.
14:26:17 <gang> I just update two patch about the readme.rst and a volume_type apischema. Please review.
14:26:50 <zhaochao> gang: yes, I get the notification email, will review them later
14:27:08 <songjian1994> https://review.openstack.org/#/c/571652/
14:27:54 <songjian1994> I discussed this with gang. Maybe we can discuss it again.
14:28:07 <gang> yes, almost forget.
14:29:09 <zhaochao> f/me doing a quick looking through on review 571652
14:29:13 * zhaochao doing a quick looking through on review 571652
14:30:22 <zhaochao> ok, this seems still another problem about datastore versioning support
14:30:32 <songjian1994> yes
14:30:54 <songjian1994> gang tends to create multi-version control, and I prefer to simply make judgments on the back-end storage itself.
14:31:57 <zhaochao> yes, it seems we don't have many better solutions before we get more capabilities to datastore version
14:32:32 <gang> and songjian and I didn't find a beautiful way to do the multi-version control.
14:33:28 <songjian1994> Yes, the version of the database that mysqlcommon manages looks a lot
14:35:55 <songjian1994> I think that such a change looks even more impressive from a more flexible perspective. Our users may have a variety of small version changes, and we can hardly take care of these。
14:35:57 <songjian1994> Although our trovestack can control the version while doing mirroring, customers seem to prefer a more liberal way
14:39:38 <zhaochao> I was thinking about datastore version capability about mapping configuration paramters to versions, though this may be a bigger change(and we could also mapping charset and collation to datastore versions too)
14:41:19 <zhaochao> datastore versions should first defined as some stricter type data in database, and we could compare them, currently datastore version behaves just like a name
14:42:29 <songjian1994> I think it is worth planning
14:43:03 <songjian1994> Configuration also have this problem
14:43:17 <gang> yes, it is worth not just for mysql
14:46:01 <zhaochao> yes, I remember the basic ideas about this kind of change have been raised before(both bp and commits were submitted), but none of them worked out(maybe only because the developer didn't work on them anymore)
14:47:28 <zhaochao> I'll also add this to my todo list
14:49:10 <songjian1994> ok,no more from me
14:50:09 <zhaochao> #action zhaochao do the basic research about datastore versioning improvement, submit a bp if possible
14:50:40 <zhaochao> ok, back to the python 3 support
14:53:10 <zhaochao> I did some basic research yesterday, in trove api (base_wsgi) we're assigning raw text values to webob response body, this is not permitted under python 3
14:54:33 <zhaochao> on the other side, troveclient, (we're still using troveclient compat part, which is based on httplib2), the reqeust response body is bytes other str under python 3
14:55:04 <zhaochao> currently these two problems are blocking the senario tests
14:56:52 <zhaochao> pylint also produce more errors under python 3, but those ones could be fixed by modifying the trove-pylint.conf, although removing trove-pylint is also suggested, but I think currently we may not have bandwith to do that
14:56:59 <gang> It is a pain from py2 to py3 0.0
14:59:08 <zhaochao> however it was decided that python 2 will be no longer supported after 2020, and many linux distribution are following this direction
15:00:36 <zhaochao> oh, didn't notice the time is up
15:00:47 <zhaochao> thanks, songjian1994 gang
15:01:09 <zhaochao> if you have time, please help to finish the python 3 first tasks
15:01:12 <zhaochao> thanks again
15:01:18 <zhaochao> Good night!
15:01:29 <gang> good night!
15:01:41 <zhaochao> #endmeeting