Wednesday, 2015-04-08

*** matt__ has joined #openstack-relmgr-office02:01
*** mattoliverau has quit IRC02:02
*** dolphm has quit IRC02:02
*** matt__ is now known as mattoliverau02:02
*** dolphm has joined #openstack-relmgr-office02:07
*** asalkeld is now known as asalkeld_afk04:32
*** asalkeld_afk is now known as asalkeld05:46
*** mikal has quit IRC08:22
*** mikal has joined #openstack-relmgr-office08:23
ttxmestery: added to RC1, blocking the merge of the liberty version bump08:54
openstackLaunchpad bug 1441501 in neutron "gate-neutron-fwaas-python27 fails with ImportError: cannot import name test_db_plugin" [Critical,New]08:54
Kiallttx: heya, just FYI - Designate has 2 open bugs we care about fixing before RC1 cuts, both in progress..09:31
*** zz_johnthetubagu is now known as johnthetubaguy09:52
*** asalkeld has left #openstack-relmgr-office11:33
mesteryttx: Ack, looks like pc_m is on this, thanks for the tip!13:06
ttxyep, looks like it's fully under control13:07
ttxdhellmann, devananda, nikhil_k: about library releases and stable branches, see
SergeyLukjanovttx, we have number of doc updates, so, I prefer to tag tomorrow14:05
SergeyLukjanovttx, all of them are very useful, not just cleanups14:06
SergeyLukjanovttx, hello :)14:06
ttxSergeyLukjanov: ack14:07
ttxW'll talk tomorrow then :)14:07
SergeyLukjanovttx, yup14:09
sdaguettx: so... requirements issue.14:20
ttxsdague: about that panic thread I just sent ?14:20
sdagueceilometermiddleware is currently required14:20
sdaguebut not in g-r14:20
sdagueoh, I didn't read that yet14:20
ttxjust posted 20 min ago14:21
sdagueyeh, not that14:21
openstackLaunchpad bug 1441655 in Ceilometer "ceilometermiddleware not in global requirements" [High,New]14:25
ttxsdague: I thought you couldn't add stuff to requirements unless it was in g-r ?14:26
devanandattx: ack, replying14:29
sdaguettx: this lives in one of the weird optional but not optional spaces14:32
sdaguebecause ceilometer injects this library into project paste pipelines for reporting, though those projects don't directly require it14:34
sdagueI only discovered it in trying to get this into devstack to address the ironic client breaks -
devanandasdague: in looking at how python-ironicclient is installed, i dont think we can cap it ... hoping you do. tldr; devstack/lib/ironic just does "pip_install python-ironicclient"14:35
sdaguedevananda: see
sdaguethere is also a juno version14:37
sdagueand probably needs to be an icehouse one, but I'll land those first14:37
devanandasdague: oh neat14:37
sdagueon the ironic break, arguably this is an unanticipated devstack bug from capping requirements14:38
devanandasdague: which break?14:38
devanandagah. i cant read14:39
sdagueall juno ironic jobs are fail14:39
devanandayeap. working on a fix to the client release I pushed yesterday to address that14:40
devanandaour client should be backwards compatible, but we aren't testing it properly. adam_g's got a patch up to make our client-src job voting, which would really help14:41
sdaguedevananda: it's already voting14:42
sdaguegate-tempest-dsvm-neutron-src-python-ironicclientSUCCESS in 53m 06s14:43
sdaguealso, to address this issue you'd need to handle the stable branch issue14:43
devanandasdague: IIRC from chatting with adam about this recently, the neutron job doesn't do the right things14:45
devanandasdague: which is why he's proposed said changes to replace it with the one in 17142614:46
sdaguedevananda: that might be14:46
sdaguehowever, that change would still not have prevented this break14:46
devanandasdague: sure. the same test need to also run on stable/ branches14:47
devanandabut right now we've got neither14:47
devanandastep 1: start testing the client from src on master. step 2: do that on stable branches. step 3: profit :)14:47
SergeyLukjanovttx, should we merge into rc-115:14
ttxSergeyLukjanov: I'd say yes15:15
SergeyLukjanovttx, oops, it's stable/juno, sorry15:15
ttxoh. I'd say yes15:16
devanandattx: what's the impact if ironic delays our RC until monday?16:52
devanandawe're chasing several bugs in localboot & whole disk image support16:53
*** johnthetubaguy is now known as zz_johnthetubagu18:30
david-lylettx, sdague, what's the policy around removing python 2.6 support in client type libraries? specifically, I'd like to remove if from  django_openstack_auth and release version 1.2.019:02
david-lyleDjango 1.7 and greater requires python >= 2.719:02
sdagueso, I'll leave that for ttx to decide, but that being said, releasing client libraries right now is kind of bad form19:05
david-lylesdague: I understand that and I've been ok with not doing it19:06
sdaguehonestly, I think that if juno and icehouse are capped correctly, there is no reason to have py26 support in anything19:06
sdagueanything new19:06
david-lyleok, that was my hope19:07
ttxdevananda: it's fine (Monday)20:16
ttxI prefer Monday rather than now and then do a RC2 next Wednesday :)20:16
devanandak :)20:18
ttxdavid-lyle: ok woith removing py26 support. Not so ok if that would be the only reason to release django_openstack_auth20:18
ttxas I'd rather avoid updating libraries now if we can20:18
david-lylettx: there are bug fixes and new feature support, but that can release after Kilo20:20
david-lyleand be consumed then20:20
david-lylenothing to block the release over20:20
david-lylethe only inconsistency is the py26 Django 1.7 issue20:21
david-lylebut the library supports 1.7 otherwise20:21
ttxright but the inconsistency doesn't prevent to run everything on 2.7 right20:21
david-lylettx: correct20:22
david-lylewe just can't merge the 1.7 global-requirements update with python 2.6 supported20:23
ttxso I don't think a bump jus tto tear py26 support out is actually useful20:23
david-lyleso we have django_openstack_auth 1.0.9 released and it supports django 1.7, but it's requirements file is <1.7 I believe20:24
david-lylemy only concern was around inconsistencies in library requirements20:25
ttxoh, so django_openstack_auth says django<1.7 and horizon says >=1.7 ?20:25
* ttx checks20:25
david-lylehorizon say >=1.4 ,<1.820:25
david-lyled-o-a says >=1.4, <1.720:26
david-lyle1.4.2 to be exact on the bottom end20:26
ttxIt's only consumed by Horizon right20:27
ttxso the bump only affects horison, which doesn't even have rc1 yet20:27
david-lylethat is correct20:27
ttxI guess doing a release now and bumping the g-r is not completely crazy20:27
david-lylethe released version will work unless someone is installing using pip and installs django 1.7, I think20:28
david-lylebut not sure how that looks in the distros20:28
ttxdavid-lyle: could you raise a thread on the ML to warn packagers that it's likely coming ? We can make the call tomorrow20:29
david-lylesure, I'll do that now, and since juno and icehouse versions of d-o-a are capped, we can safely remove 2.6, correct?20:30
ttxnext time would be good to spot the inconsistency earlier :)20:31
ttxbut then, htat's why we have such a long RC baking period20:31
david-lylettx: indeed20:31
ttxso we can recover from random fails20:31
*** SlickNik has left #openstack-relmgr-office21:05
david-lylettx: further looking and d-o-a is not capped in g-r, there would not be a change required21:21
david-lyleonly taken into account when placing the cap after release21:21

Generated by 2.14.0 by Marius Gedminas - find it at!