13:02:04 #startmeeting rpm_packaging 13:02:05 Meeting started Thu Nov 17 13:02:04 2016 UTC and is due to finish in 60 minutes. The chair is IgorYozhikov. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:02:07 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:02:09 The meeting name has been set to 'rpm_packaging' 13:02:21 o/ 13:02:30 Hi, let's spend sometime on agenda https://etherpad.openstack.org/p/openstack-rpm-packaging 13:03:14 o/ 13:03:22 #chair IgorYozhikov dirk 13:03:23 Current chairs: IgorYozhikov dirk 13:05:24 hi 13:05:36 nice 2 c u toabctl 13:05:43 ping toabctl, dirk, apevec, aplanas, IgorYozhikov, jpena, jruzicka, number80, kaslcrof 13:06:03 ok, if we are done with agenda - let's start 13:06:57 #topic - keystone https://review.openstack.org/396262 + naming of some of python projects like memcached & ldap (python-python-foo?) (IgorYozhikov) 13:07:35 I see that most of ppl are here and want to discuss keystone 13:07:47 and here I have 2 questions 13:08:02 let me start from the tail of topic 13:08:53 question is mostly for SUSE side - is it normal from perspective of naming policy that some of packages has python-python-*? 13:09:05 it can happen, yes 13:09:12 policy is python-$pypi_name 13:09:23 so if pypi_name includes python- , then it will be python-python-foo 13:09:42 in the case of python-ldap, the pypi name is ldap, so it would be python-ldap 13:10:00 (and we use pyldap anyway, so the question is mood, proper name is python-pyldap in the suse sense) 13:10:14 ah, so I'll will update my PR and uncomment strings like # Requires: {{ py2pkg('python-ldap') }} 13:10:29 well, thats wrong anyway 13:10:36 it should be py2pkg("pyldap") :-) 13:11:01 got it and what about python-memcached? 13:12:03 the pypi name is python-memcached, our suse package name is python-python-memcached 13:12:56 dirk, I can't find ldap on pypi, only https://pypi.org/simple/python-ldap/ 13:13:08 might be I'm doing something wrong? 13:13:54 IgorYozhikov: anyway, keystone uses pyldap.. 13:14:08 I got it, will build this dependency 13:14:51 so, it's become clear about naming 13:15:14 so let's discuss keystone 13:15:22 I need a feedback 13:16:06 I found that testrepository is using all available vcpu and fails ^( 13:16:21 IgorYozhikov: can you remove the -2 then? I'll give it a review round.. 13:16:25 so I set export TEST_RUN_CONCURRENCY=1 flag to run tests in 1 by 1 13:16:32 sure 13:16:46 done 13:17:37 also, is it necessary to add apache as run-time dependency? 13:18:08 or it should be done from perspective of cms side(puppet|chef|ansible|...)? 13:18:26 coincidentally we're just debating that wsgi stuff as well 13:18:38 some of our packages have it some don't 13:18:43 its not necessary from cms side 13:19:06 I guess we should take a decision and do it that way for all openstack services.. 13:19:18 toabctl suggested to do not include a sub-package like keystone-apache 13:19:33 which is going to install httpd config in right place 13:19:35 we don't add apache as a dependency either 13:19:58 yeah, hardcoding apache is definitely "wrong" in the sense that any wsgi server will do 13:20:13 ok, will remove it 13:20:28 we should just provide the wsgi app 13:20:38 then the deployer can decide what to use 13:21:07 got it, and the last question from my side - config folder 13:21:12 o/ 13:21:25 I created /etc/keystone/conf.d 13:21:25 hey number80 13:21:39 config folder does not work with wsgi afaik until now 13:21:39 and put there all conf files 13:21:41 hey number80 13:21:58 yes, I asked our keystone developer 13:22:36 he told me that it is better to store config files in expected folder like /etc/keystone 13:22:38 IgorYozhikov, keystone.conf should still be at /etc/keystone/keystone.conf 13:23:06 but I can also create a conf.d sub folder 13:23:46 well - if it doesnt work we dont need it. it needs to be handled somehow by the wsgi app 13:24:03 I try to find a solution for that. 13:24:04 anything else about keystone? just want to collect all suggestions and implement them :) 13:24:43 and finish with keystone to unblock the rest of dev core 13:24:53 not from me, will send a detailed review when I have time 13:26:10 dirk, according to your comments libxmlsec1-openssl1 - is required for unit tests 13:26:23 freezgun also required for unit tests 13:27:11 I have got it from test logs 13:28:51 I saw a couple of unmet dependencies in SUSE CI job 13:30:27 could some1 from SUSE side fix it? 13:31:41 number80, jpena do you have something to add according to current topic? 13:31:51 nope, nothing from my side 13:32:30 let's move further? 13:32:47 *nods* 13:32:58 #topic - packages reviews (https://review.openstack.org/#/q/project:openstack/rpm-packaging+status:open ) 13:33:31 dirk, do you have a chance to look into tests for mistral? 13:33:58 no, forgot about it 13:34:02 https://review.openstack.org/382196 - or I can add condition with_tests? 13:34:05 #action dirk look into test issues with mistral 13:34:06 o i c 13:34:11 thanx 13:34:14 I'll chekc it out 13:34:39 https://review.openstack.org/#/c/397618/ 13:35:44 if every1 fine with ^^^ I can merge it:) 13:37:51 https://review.openstack.org/#/c/380834/ - last week we discussed that epoch folder should be back-ported into mitaka branch 13:38:09 IgorYozhikov: too late :) 13:38:26 IgorYozhikov: ah, thats a precondition? 13:38:29 I can do that 13:38:41 yey 13:38:46 or can you? I don' tknow what the proper mitaka level epochs are 13:38:49 I can 13:39:00 going to chery-pick from newton 13:39:20 or just add a new commit in case of troubles 13:39:50 #action IgorYozhikov back-port of epoch folder into mitaka branch 13:40:51 anything else in PRs we need to pay attention now ? 13:42:08 I see that we have a number of PRs from September 13:44:27 #topic - pymod2pkg reviews (https://review.openstack.org/#/q/project:openstack/pymod2pkg+status:open ) 13:44:42 there is 1 PR here https://review.openstack.org/#/c/398912/2 13:45:49 this change is related to SUSE 13:47:03 this patch related to naming problem with python-ldap in keystone 13:47:50 kaslcrof, think so 13:48:38 #topic - renderspec reviews (https://review.openstack.org/#/q/project:openstack/renderspec+status:open ) 13:49:40 I guess that toabctl didn't have a chance to do something related to renderspec, right? 13:51:31 #topic open floor 13:52:14 we have 9 mins to discuss any other topics which are out of scope of agenda 13:53:15 * dirk is good 13:53:31 ah, well, maybe.. is there any update on the rdo gating? 13:53:40 3rd party ci I mean 13:53:50 I'm working on fixing networkx 13:54:00 number80 was faster than me :) 13:54:34 well, I'm progressing on that side, I'll have Gdal 2.x built soon 13:56:53 ok, anything else? 13:57:18 one thing 13:57:42 please proceed 13:57:51 is anybody using apache for wsgi (keystone, adoh, ceilometer,...) 13:58:00 me 13:58:08 keystone, aodh 13:58:11 eh, everybody, not anybody 13:58:19 number80, ^^ 13:58:26 yes, we are using it 13:58:36 and are you still shipping systemd files? 13:58:43 nope 13:58:49 useless 13:58:50 not for keystone 13:58:53 maybe we discussed that already and I missed it 13:59:02 but for ceilometer and aodh I think we still keep both options 13:59:09 we could still ship a systemd file with uswgi 13:59:11 no, we didn't 13:59:22 jpena, but its for developers only 13:59:38 toabctl, afraid it should be in different packages 13:59:54 or do not enable systemd unit after install 14:00:00 by default 14:00:04 IgorYozhikov, sure 14:00:21 I just want to know how others handle this wsgi switch 14:00:25 we are on time 14:00:26 #endmeeting