17:00:11 <gnuoy> #startmeeting charms 17:00:12 <openstack> Meeting started Mon Aug 22 17:00:11 2016 UTC and is due to finish in 60 minutes. The chair is gnuoy. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:15 <openstack> The meeting name has been set to 'charms' 17:00:17 <thedac> o/ 17:00:18 <beisner> o/ 17:00:33 <gnuoy> Hi all, I'll just pause a minute for any straglers 17:00:45 <tinwood> Hi! 17:00:50 <gnuoy> \o 17:01:19 <cargonza> o/ 17:01:25 <wolsen> o/ 17:01:29 <gnuoy> #topic Review ACTION points from previous meeting 17:01:32 <jamespage> o/ 17:01:43 <gnuoy> #subtopic Mascot 17:02:00 <tinwood> jamespage, do you have details on this yet? 17:02:06 <gnuoy> I don't think it was actually an action but did something happen after the last vote? 17:02:15 <jamespage> I emailed Heidi but have not heard back - 17:02:20 <gnuoy> kk 17:02:21 <jamespage> orangutang was the first choice 17:02:25 <jamespage> just chased 17:02:32 <gnuoy> #subtopic beisner Nominate thedac for charms-release group 17:02:38 <gnuoy> did that happen? 17:02:47 <jamespage> gnuoy, no I think we where waiting on your vote? 17:02:55 <gnuoy> orly 17:02:57 <beisner> sent to ml, received some +1s there. 17:03:08 <thedac> thanks, beisner 17:03:13 <gnuoy> I'll dig it out and +lots 17:03:32 <gnuoy> #topic State of Development for next Charm Release 17:03:38 <jamespage> gnuoy, added thedac to -release 17:03:44 <gnuoy> jamespage, thanks 17:03:49 <thedac> thank you 17:04:04 <wolsen> well earned thedac 17:04:05 <tinwood> yay 17:04:19 <gnuoy> Anyone got anything they'd like to highlight as a feature targetted for 16.10? 17:04:29 <beisner> woot thanks all, welcome thedac ;-) 17:04:42 <jamespage> no technically but I want to have a bit of a focus on our dev guide 17:04:53 <gnuoy> defo 17:04:54 <jamespage> I'll take an action to include the openstack-on-lxd README as a section 17:05:08 <gnuoy> #action jamespage include the openstack-on-lxd README as a section in dev guide 17:05:13 <beisner> prob worth getting nova serial-console feature dev on someone's list 17:05:15 <jamespage> beisner, are you still ok to contrib the amulet bits and pieces? I think generally documenting our CI there is OK 17:05:19 <jamespage> beisner, ah yes 17:05:36 <beisner> jamespage, yes i have an outstanding to-do to update the test section of the charm-guide, still plan to tackle that. 17:05:49 <gnuoy> The SDN cookie cutter charms *will* be ready for 16.10 17:06:05 <tinwood> once we know what shape they will be. 17:06:10 <gnuoy> ha! yes 17:06:21 <beisner> orangutan, naturally 17:06:29 <gnuoy> ok, anyone got anymore... 17:06:42 <gnuoy> #topic High Priority Bugs 17:06:56 <gnuoy> Any horrible bugs worth noting? 17:07:05 * jamespage thinks 17:07:14 <gnuoy> thedac, am I right in thinking there is some rabbit work still needs doing? 17:07:15 <thedac> glance-simplestreams-sync needs xenial updte 17:07:22 <jamespage> I had a juju 2.-0 trip hazard 17:07:29 * gnuoy wonders whether to mention mongo 17:07:38 <jamespage> https://bugs.launchpad.net/juju-core/+bug/1615601 17:07:38 <openstack> Launchpad bug 1615601 in juju-core "2.0 beta 14 + openstack: generated hostnames exceed 255 byte limit" [High,Triaged] 17:07:41 <thedac> gnuoy: possibly more stability testing for rabbit, yes 17:07:54 <jamespage> gnuoy, thedac: yeah we need to rip out all of the hosts management changes I made 17:08:04 <jamespage> I feel I should do that and clean up my own mess 17:08:08 <gnuoy> jamespage, it reminds me of the ol' directory-too-long back in pyjuju days 17:08:20 <wolsen> has anyone tested l3-ha for neutron gateway in the recent past? 17:08:21 <jamespage> oh for sockets 17:08:23 <jamespage> yah lol 17:08:28 <thedac> :) 17:08:32 <beisner> glance-simplestreams-sync is worth a separate and more detailed convo imho. when we last discussed i/p, we thought it might be good to deprecate it and add actions to the glance charm. 17:08:39 <gnuoy> wolsen, yes, it's part of our release testibng 17:08:57 <thedac> beisner: actions are an interesting idea 17:08:57 <wolsen> gnuoy: great 17:09:00 <gnuoy> beisner, yes, that was where we left it 17:09:03 <jamespage> beisner, agreed - we need todo that this week or it will not happen - I'll schedule a slot and we can discuss in #openstack-cahrms 17:09:16 <beisner> as g-s-s-s-s-s-s is currently not in ci, i think we should either knock it into shape, or work the necessary bits into the glance charm 17:09:25 <jamespage> second for me 17:09:41 <gnuoy> wolsen, I spoke to thedac and we're going to dig at it too 17:09:41 <jamespage> gnuoy, can you action me to organize a follow up for that pls? 17:09:46 <beisner> cool 17:09:48 <wolsen> gnuoy, stellar 17:09:48 <jamespage> (g-ss-s) 17:09:59 * jamespage waves a bbaqar 17:10:02 <jamespage> welcome! 17:10:12 <gnuoy> #action jamespage Discuss future of g-ss-s charm 17:10:13 <wolsen> gnuoy: I'd be more than happy to help as I'd like to understand if I did something wrong 17:10:27 <gnuoy> wolsen, sur, will keep you updated 17:10:43 <bbaqar> jamespage: thanks :) 17:10:49 <gnuoy> ok, moving on.... 17:10:57 <gnuoy> #topic requirements files in charms 17:11:24 <gnuoy> So, there was some talk of doing a global requirements thingy, is that right? 17:11:25 <beisner> hi. so osci is currently tripping over a dep of a dep of a dep of a req in charm-tools 17:11:39 <beisner> https://github.com/juju/charm-tools/issues/246 17:11:56 <beisner> which is fix-committed in charm-tools master, but will not resolve our issue until/if charm-tools cuts a new pypi release 17:12:24 <beisner> that underlying issue is solvable in another way though, which is to require setuptools version == that of xenial 17:12:30 <tinwood> if it's in a venv, why's it failing? is it using system packages? 17:13:02 <beisner> no, it's setuptools 2.2 in a venv on trusty vs setuptools 25.2.0 in venvs on xenial. 17:13:36 <jamespage> beisner, is this just effecting UOSCI testing? is OpenStack verification ok? 17:13:38 <beisner> if we use the later setuptools, problem solved. 17:13:56 <beisner> jamespage, openstack verification is ok, as they're now building venvs on xenial hosts. 17:14:00 <tinwood> I think it is using site-packages. However, it may need to. 17:14:37 <beisner> on the topic of xenial, we will need jenkins-slave to overcome this https://bugs.launchpad.net/charms/+source/jenkins-slave/+bug/1543380 17:14:38 <openstack> Launchpad bug 1543380 in jenkins (Ubuntu) "jenkins-slave on Xenial: Unable to connect to Upstart: Failed to connect to socket " [Undecided,Confirmed] 17:14:52 <beisner> back to the nature of the topic, and my question: 17:15:13 <beisner> requirements.txt and test-requirements.txt across os-charms are not unified 17:15:15 <beisner> but should be imho 17:15:37 <beisner> i've added links to upstream requirements-update code, which may be usable for this, albeit overkill for what i'd like to do. 17:16:08 <beisner> which is maintain requirements in our release-tools repo, along with the necessary script to push that into the os charm repos. 17:16:18 <gnuoy> sorry, where have you added those links? 17:16:27 <beisner> oh sorry, in the etherpad 17:16:30 <beisner> https://etherpad.openstack.org/p/openstack-charms-weekly-meeting-20160822 17:16:52 <gnuoy> #link https://github.com/openstack/requirements 17:16:59 <gnuoy> #link http://docs.openstack.org/developer/requirements/#enforcement-in-projects 17:17:06 <beisner> jamespage, without charm-tools passing, we won't get amulet tests because it's bail on first fail in the pipeline 17:17:24 <jamespage> beisner, yeah 17:18:14 <beisner> so, fact: charm-tools in a venv on a trusty host now fails by default until pypi pkg is revved up; qq: do we pressure there and wait it out, or start modding charm reqs? 17:18:44 <jamespage> beisner, hmm 17:19:09 <beisner> it does bring the desire to have global reqs swiftly to the front burner, if that's the direction we want to go. 17:19:23 <jamespage> my main problem is that even if we go for an approach that means we have consistent requirements.txt and test-requirements.txt we can still get hit by transient dependency issues 17:19:57 <beisner> yeah, in this case the dependency chain looks like: cryptography->secretstorage->keyring->launchpadlib->charm-tools 17:19:58 <jamespage> I think its a step forward, but we should probably follow our peer projects and start to use pip constraints to pin/fix versions as required 17:20:22 <beisner> where the breakage was at charm-tools, but on a launchpadlib dep, when secretstorage revved up. 17:20:32 <beisner> i'm not sure anything can solve that 17:20:44 <jamespage> the constraints are sourced from upper-constraints.txt in global-requirements - if we take that approach, then we can fix a transient break in a central way, rather than having to raise 30 reviews 17:21:00 <beisner> yes that is good 17:21:17 <gnuoy> ok, I think we're agreed on the central requirements approach 17:21:23 <jamespage> agreed +1 17:21:26 <beisner> but if you take this breakage, we would have to pin 4 layers deepo 17:21:29 <beisner> deep, even 17:21:35 <gnuoy> beisner, but in one place 17:21:39 <tinwood> I'm not sure I understand it. 17:21:42 <jamespage> beisner, we can generate a constraints file 17:21:55 <beisner> cryptography->secretstorage->keyring->launchpadlib->charm-tools 17:22:07 <jamespage> based on a known good baseline, and then refresh and re-test on a regular basis 17:22:11 <beisner> ^ secretstorage revved up, breaking charm-tools. we already pin charm-tools. 17:22:52 <jamespage> beisner, I think constraints covers everything in the venv 17:22:52 <jamespage> https://github.com/openstack/requirements/blob/master/upper-constraints.txt 17:23:06 <jamespage> not just the expressed dependencies in requirements.txt 17:23:26 <beisner> right, i don't disagree there at all. ok, so who's going to take that on? 17:23:52 <beisner> seems like a non-trivial undertaking 17:24:02 <jamespage> I don't think its that bad 17:24:10 <jamespage> we have a good template to copy imho 17:24:42 <jamespage> I know everyone is super busy 17:24:54 <beisner> i would like to get there (use openstack/requirements as a model). i would also like to unblock your gate tactically. 17:25:07 <jamespage> beisner, quickest way to gate being unblocked is to release charm-tools right? 17:25:11 <beisner> yes 17:25:19 <jamespage> lemme go kick the tyres there then 17:25:35 <beisner> fyi, queried in #juju first thing this am 17:25:38 <beisner> worth a re-ping 17:26:18 <jamespage> beisner, ok 17:26:22 <gnuoy> ok, so publish fixed charm-tools RSN and global requirements soon too 17:26:30 <jamespage> gnuoy, agreed 17:26:36 <gnuoy> kk 17:26:42 <gnuoy> #topic Barcelona Design Summit: Fishbowls, Workrooms and Contributor meetups. 17:26:48 <gnuoy> jamespage, your topic I believe 17:27:05 <jamespage> oh yeah - so we need to decide on what to request for the design summit in terms of slots etc... 17:27:15 <gnuoy> jamespage, whats the deadline? 17:27:20 <jamespage> 31st 17:27:27 <gnuoy> ack 17:27:31 <jamespage> fishbowls are fixed advertised content - I was going to suggest 17:27:53 <jamespage> 1) retry of 16.07 and 16.10 charm releases - identify pain points and potential ways forward 17:27:58 <jamespage> retro 17:28:16 <jamespage> 2) 16.10 planning and high level objective setting 17:28:35 <jamespage> and then request a workroom for an afternoon (3 slots) 17:28:49 <jamespage> with a contributor meeting on the friday pm if we can get one 17:28:53 <jamespage> does that sound ok? 17:28:59 <gnuoy> sounds good to me 17:29:02 <thedac> sounds good 17:29:03 <tinwood> Sounds excellent. 17:29:26 <jamespage> okwill request tommorow 17:29:42 <gnuoy> #action jamespage request Barcelona Design Summit slots 17:29:53 <gnuoy> we're pretty much out of time 17:30:07 <gnuoy> anyone got anything else burning? 17:30:16 <thedac> no 17:30:18 <tinwood> nope 17:30:24 <jamespage> no 17:30:26 <gnuoy> same time, same place in 2 weeks then 17:30:32 <gnuoy> #endmeeting