13:02:45 #startmeeting rpm_packaging 13:02:46 Meeting started Thu Jan 12 13:02:45 2017 UTC and is due to finish in 60 minutes. The chair is IgorYozhikov. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:02:47 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:02:49 The meeting name has been set to 'rpm_packaging' 13:03:07 #chair jpena dirk IgorYozhikov 13:03:08 Current chairs: IgorYozhikov dirk jpena 13:03:29 o/ 13:03:41 o/ 13:03:46 let's spend some time on agenda https://etherpad.openstack.org/p/openstack-rpm-packaging 13:03:56 o/ 13:04:16 #chair jpena dirk IgorYozhikov number80 13:04:17 Current chairs: IgorYozhikov dirk jpena number80 13:04:36 o/ 13:06:32 let's start 13:06:39 yep 13:07:04 #topic keystone https://review.openstack.org/#/c/396262/, let's merge it? 13:07:22 if the pymysql change is done, I'm good 13:07:29 I tried to address all comments from all sides in my last PR 13:07:57 number80, I just updated PR with pymysql 13:08:30 but now it failed on SUSE ci side 13:09:10 I think the eventlet package in suse CI is missing a deps if I stand correctly 13:09:51 dirk, could you take a look please ^^^^^ 13:12:00 so, does it means that in common everything else is fine with keystone PR? 13:13:04 and it is only required to fix possible issues related to SUSE ci passing 13:13:09 and we can merge 13:13:14 right? 13:15:20 number80, dirk, jpena your opinion? ^^ 13:15:43 yes, I'm fine with that (I'd like to fix the ldap/pyldap issue, but that's the next topic) 13:15:51 IgorYozhikov: I need to review it and fix the suse ci issue 13:16:01 great, thanx! 13:16:14 #topic pyldap or python-ldap, change mappings in pymod2pkg or just use exact component from project requirements 13:16:54 I'm not sure that we need to create one mapping for 2 different projects 13:17:22 let's discuss this topic, question raised by apevec 13:17:23 we can handle that with provides 13:18:43 as dirk mentioned in the keystone code there is pyldap 13:19:11 number80, do you suggest to add provides pyldap into python-ldap, am I right? 13:19:17 number80: the point is that keystone explictely requests pyldap 13:19:26 so pyldap needs ot be the eggname 13:19:29 ldap is not the same thing 13:19:38 agree 13:21:45 just trying to understand why not to use pyldap. Is there any blockers on the RH side for that? 13:22:14 * number80 trying to remember why 13:23:22 I need to recheck with Alan, but we can provide pyldap somewhere 13:23:45 in the fedora packages, we have python-ldap and python3-pyldap (pyldap supports python3, but only that subpackage is created) 13:23:47 it could be a bug or something 13:24:28 jpena: python2 variant can be reenabled 13:25:04 but it means phasing out legacy ldap python modue 13:25:37 (ok, just remembered that was the problem) 13:27:19 ok, so number80 is going to clarify usage of pyldap with apevec, right? 13:27:29 ack 13:27:39 #action number80 check with apevec pyldap status 13:27:45 cool 13:28:15 but I lean toward switching to pyldap either for ocata or pike at last 13:29:50 #topic Open Floor 13:30:43 I want to say welcome to new contributors - Dobroslaw, kornicameister 13:30:54 sorry if missed someone 13:31:01 and jakub :) 13:31:10 sure, thanx 13:31:33 hi all 13:31:54 I think that you may have questions || topics to discuss 13:32:17 Welcome Droboslaw and kornicameister 13:34:09 I guess there are some questions from Dobroslaw and kornicameister on the openstack-rpm-packaging channel 13:34:48 [07:36] I'm working now on monasca-api in https://review.openstack.org/#/c/418376/ 13:34:54 [07:36] And I have in some way similar problem with dependencies like kornicameister 13:35:04 [07:37] in monasca-api requirements file we have 3 potential db drivers, but two of them are commented: https://github.com/openstack/monasca-api/blob/master/requirements.txt#L20 13:35:09 [07:38] and because of that tests fail (not seeing casandra driver) 13:35:14 [07:38] should I add all this drivers as dependencies in spec file? 13:35:23 [07:39] user will most likely need only one specific - for him - driver 13:35:27 [07:39] but tests needs all 13:36:04 (sorry for putting it like that, but Dobroslaw had to leave) 13:36:54 jakub: I thought I answered to that 13:37:04 ok I have some more time 13:37:17 dirk> kornicameister: generally sqlalchemy 13:37:17 07:12 So just depend on that .. the actual orm is an installer issue 13:37:17 07:12 Usually OpenStack uses oslo.db though 13:37:34 yes, but for relational dbs 13:38:11 I'm looking for Casandra, InfluxDB and Vertica 13:38:43 ah, so you're talking about the test failures? 13:38:45 just add it as buildrequires 13:38:47 yes 13:38:52 buildrequires are not runtime requires, so it doesn't affect users 13:39:05 and about drivers for end users 13:39:23 they will use one driver 13:39:36 do you want to opinionate to influxdb? 13:39:38 but how to pack it? 13:40:00 make this all drivers dependency of one package? 13:40:01 are you speaking about https://github.com/openstack/monasca-api/blob/1.5.0/tox.ini#L14-L18 13:40:52 or create 3 packages, every one depending on one driver? 13:41:16 IgorYozhikov, well partially, tests check all drivers 13:41:31 in these cases, we usually leave the installation of the driver as a task for the deployer 13:41:39 as runtime req I mean 13:42:14 Dobroslaw: well, either opinionate one or none 13:42:19 Dobroslaw: IMHO I would not require any 13:43:00 Dobroslaw: its the task of the installer (manualler or automatic) to setup the database 13:43:02 so running installation of influxdb from e.g. chef based on user config? 13:43:13 sure, I mean the database could be on a completely different node 13:43:21 so requiring any database doesn't make sense 13:43:24 if these packages are required only during build time you can add them to build-requires or exclude such tests from testr 13:43:31 sorry, I mean influx python driver 13:44:40 Dobroslaw, does it means that obly drivers will be required for tests, not DBs, right? 13:44:47 yes 13:45:00 same thing, if you don#t want to opinionate the pacakging to always use influxdb, just depend runtime wise on nothing 13:45:54 oh, ok 13:45:58 will do 13:46:00 thank you 13:46:04 yep 13:47:28 anything else to discuss - topics, questions, concerns? 13:47:36 yes 13:47:53 #link https://releases.openstack.org/ocata/schedule.html 13:48:37 schedule reminder 13:48:46 R-5? 13:48:59 feature freeze is upcoming, and PTL self nomination next week 13:49:08 In case you want to think about becoming PTL :) 13:49:24 o i c :) 13:49:27 sorry, week after next week 13:49:30 I'm off by one 13:50:22 yep, resigning from this cycle, so feel free to candidate :) 13:51:48 number80, this cycle looks to be shorter compare to newton 13:52:24 anyway, thank you for info & reminding 13:52:40 well, it's nice to change from time to time 13:53:34 number80, what is about ptg? any updates? 13:54:40 as I remember that we planed to create remote sessions if it possible 13:55:28 Well, we have 2 days scheduled of meeting, and yes, we'll plan remote sessions (likely during the US mornings/European afternoons) 13:55:48 yey :D 13:55:59 I haven't received exact schedule to make suggestions yet 13:57:03 good, anything else? 13:58:18 * dirk is good 13:58:36 * jpena to 13:58:38 too 14:00:03 #endmeeting