Thursday, 2018-03-15

openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer-powervm master: Updated from global requirements
openstackgerritOpenStack Proposal Bot proposed openstack/networking-powervm master: Updated from global requirements
openstackgerritOpenStack Proposal Bot proposed openstack/nova-powervm master: Updated from global requirements
*** AlexeyAbashkin has joined #openstack-powervm08:13
*** adi______ has joined #openstack-powervm11:20
*** Alexey_Abashkin has joined #openstack-powervm11:20
*** adi_____ has quit IRC11:27
*** adi______ is now known as adi_____11:27
*** AlexeyAbashkin has quit IRC11:28
*** adi_____ has quit IRC11:28
*** adi_____ has joined #openstack-powervm11:28
*** Alexey_Abashkin is now known as AlexeyAbashkin11:28
*** edmondsw has joined #openstack-powervm12:15
*** apearson has joined #openstack-powervm12:59
*** tjakobs has joined #openstack-powervm13:53
*** esberglu has joined #openstack-powervm14:00
-openstackstatus- NOTICE: POST_FAILURE results on Tempest-based jobs since the most recent Zuul security fixes are being investigated; rechecking those won't help for now but we'll keep you posted once a solution is identified14:10
*** ChanServ changes topic to "POST_FAILURE results on Tempest-based jobs since the most recent Zuul security fixes are being investigated; rechecking those won't help for now but we'll keep you posted once a solution is identified"14:10
openstackgerritMerged openstack/nova-powervm master: Updated from global requirements
edmondswesberglu see thread starting with
efriedesberglu: Can you look into this failure when you get a chance please:
esbergluefried: That's what the thread ^ is about14:22
efriedoh, okay.14:22
efriedI leave it in your capable hands.14:22
edmondswthis is also going to be interesting:
efriededmondsw: It'll make it easier for us to bump pypowervm OOT.  Otherwise, I don't really see a huge impact for us.14:31
esbergluefried: edmondsw: I'm getting this error on most of the UTs after following the info in that thread14:55
esberglu^ seems like we need to update that14:55
efrieduh oh14:55
efriedThis may have something to do with the big "policy in code" movement that peaked in Queens.14:56
efriedIf you get stuck, ask lbragstad in -keystone14:56
edmondswI can help with that14:56
edmondswlet me look14:56
efriedOr ask edmondsw.14:56
esbergluedmondsw: Only changes I have made so far14:57
esberglu- remove neutron from deps in tox.ini14:57
esberglu- add neutron to requirements.txt14:57
esbergluthat's it14:57
edmondswesberglu when we were using git, etc/policy.json would be wherever the git dir was, so this would work15:02
edmondswbut pypi isn't going to install that file into site-packages15:03
edmondswI'm not sure if it would put it in /etc/policy.json or just not install one at all15:03
edmondswprobably the former15:03
edmondswI would try changing that method to look in /etc/policy.json instead of trying to build a path based on base.__file__15:04
edmondswthen to efried's point we probably also need to handle the case that the file doesn't exist... but may not have to solve that to get this working15:06
edmondswlet me try it15:06
*** apearson_ has joined #openstack-powervm15:31
*** apearson has quit IRC15:33
openstackgerritEric Berglund proposed openstack/networking-powervm master: Move neutron to requirements.txt
esbergluedmondsw: efried: ^ should fix tox16:10
openstackgerritEric Fried proposed openstack/networking-powervm master: Updated from global requirements
efriedesberglu: Thanks.  Rebased ^ on top of it.16:14
*** AlexeyAbashkin has quit IRC16:32
edmondswesberglu yep, there is a second version of the file there, and appears to be identical. interesting17:11
edmondswesberglu sorry for the delay. Ben dropped in17:11
openstackgerritMerged openstack/networking-powervm master: Move neutron to requirements.txt
efriededmondsw: want to pull out of the gate til it passes our CI?17:26
edmondswefried it had already passed our CI on an earlier run... still think I should?17:26
efriedI yanked it.17:27
efriedand yes, definitely.17:27
efriedcause now it's based on something else that, who knows, could affect it.17:27
edmondswI guess... not sure how... but done17:27
edmondswefried you can remove your -2 now17:28
edmondswI added a CI recheck17:28
efriedActually don't know how the gate works on that.17:28
edmondswefried oh, because the gate jobs had already started... yeah, might need to keep it, I dunno17:28
efriedIf it already went into the gate... right17:28
edmondswI would hope not, but better safe I guess17:29
efriedI don't know either, so I'll just leave it in place for now.17:29
edmondswefried gate succeeded, so should be ok to remove the -2 now17:38
*** AlexeyAbashkin has joined #openstack-powervm18:54
*** AlexeyAbashkin has quit IRC18:58
*** esberglu has quit IRC19:07
*** esberglu has joined #openstack-powervm19:09
*** esberglu_ has joined #openstack-powervm19:12
openstackgerritMerged openstack/networking-powervm master: Updated from global requirements
*** esberglu has quit IRC19:14
*** ChanServ changes topic to "This channel is for PowerVM-related development and discussion. For general OpenStack support, please use #openstack."19:25
-openstackstatus- NOTICE: The regression stemming from one of yesterday's Zuul security fixes has been rectified, and Devstack/Tempest jobs which encountered POST_FAILURE results over the past 24 hours can safely be rechecked now19:25
edmondswesberglu we're not running CI on neutron or ceilometer changes, only neutron-powervm and ceilometer-powervm changes?19:33
edmondswremind me why?19:33
esberglu_edmondsw: We actually do run on neutron/ceilometer, we just don't vote19:34
esberglu_We would need to talk to those projects individually to get approved19:35
edmondswyou mean we don't publish results19:35
edmondswok, I'll add a TODO for that19:35
esberglu_There are also neutron and telemetry specific tempest plugins that we would probably have to be running in addition to or instead of base tempest19:36
esberglu_I haven't really looked into those much19:36
edmondswoh, we do have that on the TODO list :)19:38
edmondswadded the tempest plugin bit... didn't see that there19:39
edmondswlow priority19:39
esberglu_edmondsw: efried: Is resize not supported if using ssp?20:25
esberglu_Thought I saw some tempest tests hitting that, but I don't think it is20:25
efrieda resize that includes disk, or any resize?20:26
efriedLet me unpack my brain on you:20:26
edmondswesberglu_ I believe resize is *only* supported if you are using ssp20:26
esberglu_Looks like it doesn't work if you are resizing to a larger flavor20:27
efriedI do recall that you can't resize an LU.  If you do that, you end up with an LU of the new requested size... full of zeros.20:27
edmondswlarger in what sense?20:27
edmondswlarger disk?20:27
esberglu_Larger disk, ssp doesn't have an extend_disk() method20:27
efriedSo if we were to support such resize, we would have to create a new LU and copy the data over.20:27
edmondswah... that may be20:27
edmondswlarger CPU or mem should be fine20:27
efriedOr, yeah, convince the platform to support extend_disk.20:27
efriedSo right, knowing that, we may have blocked resize of SSP-backed VMs entirely.20:28
efriedwe might have been smarter and allowed resizes as long as they don't change disk size.20:28
edmondswif we wanted to create a new LU and copy over, the good news there is that community resize is not live resize, which vastly simplifies that20:28
efriedAnd I don't know offhand which we did.20:28
esberglu_Looks like resize works for SSP only, but not if you want to extend the disk20:29
esberglu_cool tx20:29
edmondswlooks like it would actually try to ExtendDisk... but we're saying that won't work?20:31
edmondswkinda silly to try it then20:31
esberglu_edmondsw: It would give you a not implemented error if you tried it20:31
edmondswI guess that's there for the _is_booted_from_volume case20:32
esberglu_But we could probably just raise the NotImplementedError as soon as we see the new size is bigger20:32
edmondswbut I don't think we support bfv for anything other than PowerVC?20:33
edmondswefried do you know if bfv would work with nova-powervm and a non-PowerVC OpenStack (e.g. devstack)?20:33
edmondswwonder if anyone has ever tried that20:34
efriedI don't know.  AFAIK, BFV is new in Q, so the answer is probably "we haven't tested it, no idea"20:34
edmondswI don't think that's new in Q...20:34
edmondswe.g. this is newton:
efriedHum, we have code around is_booted_from_volume.20:36
edmondswall PowerVC VMs use a volume as the boot disk... but that's not what nova calls bfv20:36
edmondswI wonder if that's what that method is covering today, though20:37
edmondswwe may have to try it and see20:37
edmondswthorst may have some knowledge he can drop on us20:37
edmondswesberglu_ I would recommend trying bfv with the in-tree vSCSI commit, and if that doesn't work we can simplify the code as we port resize in-tree20:41
edmondswif it does work, then we need to keep the ExtendDisk for the bfv case20:41
*** edmondsw has quit IRC21:02
*** apearson_ has quit IRC21:25
*** esberglu_ has quit IRC21:41
*** esberglu has joined #openstack-powervm21:44
*** esberglu_ has joined #openstack-powervm21:45
*** esberglu has quit IRC21:48
*** esberglu_ has quit IRC21:50
*** tjakobs has quit IRC21:59

Generated by 2.15.3 by Marius Gedminas - find it at!