Tuesday, 2017-01-10

*** jwcroppe has quit IRC00:01
*** jwcroppe has joined #openstack-powervm00:01
*** edmondsw has quit IRC00:02
*** jwcroppe has quit IRC00:06
*** jwcroppe has joined #openstack-powervm00:13
*** jwcroppe has quit IRC00:19
*** jwcroppe has joined #openstack-powervm00:31
*** thorst has joined #openstack-powervm01:01
*** svenkat has joined #openstack-powervm01:12
*** thorst has quit IRC01:17
*** thorst has joined #openstack-powervm01:20
*** thorst has quit IRC01:32
*** tjakobs has joined #openstack-powervm01:37
*** tjakobs has quit IRC01:44
*** svenkat has quit IRC02:45
*** edmondsw has joined #openstack-powervm03:34
*** edmondsw has quit IRC03:38
*** AlexeyAbashkin has joined #openstack-powervm05:27
*** AlexeyAbashkin has quit IRC06:24
*** AlexeyAbashkin has joined #openstack-powervm07:09
*** k0da has joined #openstack-powervm07:59
*** BorD__ has quit IRC09:38
*** openstack has joined #openstack-powervm14:28
*** smatzek has quit IRC14:28
adreznecI don't think I'll have the bandwidth to make those kind of changes before next week14:29
thorstso when esberglu gets back next week...his priorities are integrating your new server, then a power CI14:29
thorstand he'll work with you on it?14:29
adreznecnew server?14:29
*** wangqwsh has joined #openstack-powervm14:29
thorstthat DNS thing14:29
adreznecAh, right14:29
thorstand apt-cache14:29
efriedthat sounds like lower priority.  is it?14:29
thorstlower priority, sure...but it has the potential to shave 20-30 minutes off each run14:30
thorstand is now minimal effort to add in14:30
adreznecWell other than the fact that it should just be picked up in a redeploy14:30
thorsta day or so14:30
thorst(cause adreznec did all the real work)14:30
adreznecSo yeah, I guess we're stuck polishing until next week14:30
thorstbut we're in agreement on how to move forward.14:31
adreznecOkay then14:31
adreznecLet's see14:32
adreznec#topic CI Status14:32
adreznecSeems like things aren't doing too badly14:32
thorstadreznec: we definitely have some failures that aren't the patch14:32
adreznecWe're doing worse than KVM14:32
thorstbut generally....yeah, pretty good14:33
adreznecBut better than hyper-v14:33
thorstI think that can be a lower priority at the moment.14:33
adreznecDefinitely room for improvement but yeah, not top priority14:33
thorstongoing optimization effort14:33
efriedkvm aka the golden child14:33
thorstand I wonder if we'll be doing better once we have that other server in place14:33
efried<rolls eyes>14:33
thorstto help reduce external dependencies14:33
adreznecWell that's the baseline, like it or not :P14:33
efriedteacher's pet14:34
adreznecefried: Pretty much14:34
adreznecSo yeah, I think unless something unravels this week we'll try and maintain status quo until esberglu returns14:34
adreznecand we have our next steps for when he returns14:35
adreznecAnything else on CI?14:35
efried#action esberglu to publish CI results in Thai14:35
adreznec#topic Ocata testing14:36
adreznecSo I know we've been a bit fragmented this cycle, with a good chunk of active test (at least from the sde side) happening still on Newton14:37
adreznecDo we feel like we have any gaps in new function that we should be testing with?14:37
*** tblakes has quit IRC14:38
adreznecthorst: Do you know if the neutron guys get anywhere addressing that OvS live migration downtime issue this cycle?14:38
adreznecotherwise from Nova it's mostly resource providers and placement stuff, which should be above us14:38
thorstadreznec: I haven't seen anything  :-(14:38
thorstbut I haven't done a good deal of research yet14:39
adreznecWe're basically 2 weeks out from O-314:40
adreznecSo we'll definitely want to identify anything we think we need fixes with14:40
efriedDo we have anything we need to get into ocata?14:40
efriedYeah, what he said14:40
thorstI don't know of anything at the moment14:40
thorstthe GPFS cinder volume driver integration will likely go in pike14:41
adreznecWhat about the SSP vol adapter14:41
thorstneed to ask Shyama actually about that14:41
thorstand then the image cache...14:41
adreznectjakobs is still working on that14:42
adreznecI think we're on track to get that in for sde mode14:42
adreznecVIOS bug there might keep it from getting in ocata for VIOSes though14:42
thorstadreznec: how to tell if true VIOS or SDE VIOS14:42
thorstI guess he can check that by looking at the is_mgmt_partition bit14:43
adreznecis management partition14:43
thorstgood to know before I review that14:43
adreznecIt's not that he's writing the code in a way to preclude it14:43
adreznecBut he's having issues with cplv on VIOS and has put a defect in14:43
adreznec(he was out yesterday so not sure what the status of that is)14:44
thorstadreznec: OK - so just have him bug us when he's ready for reviews14:44
thorstI'm so heads down on that gpfs cinder thing that I have neglected many other things14:44
adreznecOther than that I don't think there's anything else I'm aware of we're trying to get into Ocata from new function14:45
adreznecand CI doesn't seem to have pointed out any existing function we're broken with14:45
adreznecThat we haven't already fixed, anyway14:45
* adreznec looks at you, async power on job14:46
adreznecSo yeah. The only thing I'm not really sure how much testing we've gotten in on is the ceilometer-powervm support14:46
adreznecI've been using OSA almost exclusively without Ceilometer due to the fact that it blows up the disk on my environments14:47
*** k0da has joined #openstack-powervm14:47
thorstadreznec: we should just kick off a patch14:47
thorstmaybe update the requirements14:47
thorstbut I do think it gets installed in the CI14:47
adreznecYeah, it does14:47
adreznecBut yeah, a reqs bump to match g-r is probably in order14:48
thorstacross all the projects...14:48
adreznecI'll toss that on my backlog unless someone else wants to pick it up14:48
adreznecEerily quiet14:49
adreznecAll right14:49
adreznec#topic Open flooor14:49
adreznecAnything else people have?14:49
thorstjust an idea on how we can decide future design decisions14:50
thorstseroyer and I were talking...when we have a binary decision like 'lets use X instead of Y'14:50
thorstrather than have a technical discussion on it14:50
thorstlets play overwatch instead, and if Team 1 wins, we choose X, and if Team 2 wins, we choose Y14:50
adreznecThat seems reasonable14:51
adreznecIf we're pressed for time, we can just pick a random twitch stream and skip to the end to see which team wins14:52
adreznecOne other thing I've been thinking of14:53
adreznecWhen we land the base powervm driver in-tree (either in Ocata or Pike)14:54
adreznecDo we want to fold ceilometer-powervm into ceilometer proper?14:54
thorstadreznec: culture of ceilometer is that it is folded in14:54
adreznecSince it should be usable with both14:54
efriedDo we need a separate blueprint for that?14:55
adreznecIf we're good with that then I guess we can just leave it as-is14:55
thorstI'd say leave it for now14:55
adreznecWas just thinking it's one less project to maintain14:55
*** dwayne__ has joined #openstack-powervm14:55
adreznecFor requirements, etc14:55
thorstfully agree...14:56
adreznecBut we can re-address down the road14:56
thorstbut they were moving towards a drivers should be out of tree14:56
thorstat least they were when we started14:56
adreznecthat's all I had14:57
thorstI have a pypowervm q to efried14:57
thorstdo you know where we are with that patch that doesn't destroy a VIOS if it is hosted via another VIOS14:57
adreznecmeeting over, back to your regularly scheduled channel14:57
efriedthorst: bwarrum was working on that, but I haven't seen progress in a couple days.14:58
efriedI'll circle back with him.14:58
thorstefried: thx15:00
*** smatzek has joined #openstack-powervm15:03
*** tblakes has joined #openstack-powervm15:05
*** wangqwsh has quit IRC15:11
*** AlexeyAbashkin has quit IRC15:15
*** AlexeyAbashkin has joined #openstack-powervm15:16
*** AlexeyAbashkin has quit IRC15:19
*** kriskend_ has joined #openstack-powervm15:24
*** burgerk has quit IRC15:37
*** burgerk has joined #openstack-powervm15:45
*** tjakobs has joined #openstack-powervm15:58
efriedadreznec, thorst: I'd like to stack a system to test out the in-tree driver patches.  Before I go down one or more of the ratholes I'm staring at, do y'all have any tips for what my local.conf should look like?16:05
thorstefried: I do not...I suspect similar but do not enable the out of tree driver bits in it.16:06
thorstand nova-compute might fail to start unless you get your patch pulled down prior to stacking16:06
adreznecthorst: Yeah, that's no different than any other in-dev patch though16:07
efriedFirst patch has no network and no disk.  So... will the spawn even "succeed"?16:07
efriedI guess we'll power on, but the partition will sit in SMS/OF because no disk16:08
adreznecI think it will still spawn though16:09
efriedSo do I just omit all the local.conf cruft about networking completely?16:09
efriedOr will I still get an IP, just no plug?16:09
adreznecWell... if this system is in sde mode/using ovs I think that should be the default16:10
thorstyeah, I'd toss out the local.conf stuff about networking16:10
thorstsomeone needs to learn pastebin....16:10
thorstthat just leave in...it won't hurt leaving it in, may break devstack taking it out16:10
thorstbut the part where you enable pvm-sea or networking-powervm, you could probably take that out16:11
adreznecWell I mean most of those things should have defaults16:11
thorst(though again, maybe it doesn't hurt leaving it in)16:11
adreznecSo they shouldn't be needed16:11
adreznecSince we're not doing anything with it anyway16:11
efriedNot an SDE system, btw.  Should it be?16:12
adreznecWell it could be SDN mode, but I don't think we're planning to support SEA networking initially... so it should definitely be running OVS16:12
adreznecUnless I'm wrong thorst16:12
adreznecIf it's already a Novalink+VIOS setup it's probably easiest to test using Novalink w/ OVS networking and VIOS w/ localdisk storage I think16:13
efriedFirst patch doesn't have any disk or networking.16:15
adreznecSure first patch it won't matter, but if you set it up it should work for testing subsequent patches in the series16:15
adreznecand you can get away with minimal devstack configuration16:16
thorstI'm not sure on the OVS or SEA path initially16:19
thorstpart of it will be how the os-vif stuff goes16:19
thorstbut I wouldn't worry about it atm16:19
thorstbaby steps16:19
*** burgerk has quit IRC16:24
*** burgerk has joined #openstack-powervm16:30
*** burgerk has quit IRC16:30
*** tblakes has quit IRC16:43
*** jwcroppe_ has joined #openstack-powervm16:44
*** apearson has quit IRC16:45
*** jwcroppe has quit IRC16:45
*** dwayne__ has quit IRC16:45
*** tblakes has joined #openstack-powervm16:49
*** apearson has joined #openstack-powervm16:49
*** dwayne__ has joined #openstack-powervm16:52
*** tblakes_ has joined #openstack-powervm16:53
*** tblakes has quit IRC16:54
*** tblakes_ is now known as tblakes16:54
*** jwcroppe_ has quit IRC16:55
*** jwcroppe has joined #openstack-powervm17:06
efriedadreznec: thorst: where are we on https://review.openstack.org/#/c/413736/ ?17:14
thorstI need to full review...I just wanted it passing CI first17:18
thorstwhich it did as of yesterday17:18
thorstso I guess its time to ... review17:18
*** tblakes has quit IRC17:20
*** tlian has quit IRC17:20
*** kriskend_ has quit IRC17:20
thorstefried: that's really it???17:21
thorstI can't believe we don't have more changes...17:21
*** kriskend_ has joined #openstack-powervm17:21
efriedthorst, it's because we've got the namespace shim in place to make it part of nova.virt17:21
efriedWe don't need to change any of the nova_powervm.virt namespace.17:21
thorstahh, so we leave nova_powervm.virt.powervm17:22
thorstand the shim is all that changes17:22
efriedSilver lining to the scramble from the Austin summit ;-)17:22
thorstI'm kind of shocked that we don't have any CI impacts17:22
thorstI thought for sure that would blow up17:22
efriedAll that had to change there was the `compute_driver = powervm_ext.driver.PowerVMDriver`17:23
efriedvia devstack/settings17:23
efriedand setup.cfg17:24
thorstyeah yeah...I'm just trying to figure out why that didn't blow up17:24
thorstand I see why now17:24
thorstour CI just says 'enable_plugin nova_powervm xyz'17:24
thorstwhich then pulls down the devstack/settings17:24
thorstgot it17:24
*** tblakes has joined #openstack-powervm17:31
thorstI put a -117:32
thorstsmall nit17:32
thorstthen I'm +2, but I think adreznec should also do a review17:32
efriedthorst, I'm not clear what you're suggesting - just change the title of that section to "Upgrade Considerations"?17:32
thorstor make a document called 'UPGRADE.rst' or something17:33
efriedokay.  And the rewrite... what are you looking for there?17:33
thorstnothing, that's more a note to self17:33
thorstabout that whole doc17:33
thorstbut that doc as is...I don't know that its a good readme for new users coming in17:34
thorstit may be...I'm not sure17:34
openstackgerritEric Fried proposed openstack/nova-powervm: Change namespace to nova.virt.powervm_ext  https://review.openstack.org/41373617:34
efriedI have yet to push anything to the mailing list.  Do you want me to get set up to do that, or does one of y'all want to take care of it?17:35
thorstefried: you cool doing a mailing list note on this?  I doubt anyone will respond...but its the nice thing to do17:35
thorstefried: I think you should, given it'll be more common for you with the in-tree driver focus17:35
efriedI don't mind doing it in theory, but I'm led to understand there's all kinds of setup and protocol necessary.17:35
thorstbasically, lotus notes is evil17:36
efriedWell, as I understand it, one shouldn't mail from one's work email anyway.17:36
thorstI don't think that matters...17:36
thorstmost people set up a LTC e-mail id17:36
efriedThey're always saying to create a personal account so you can keep your reputation/personality if you change companies and continue to work in the community.17:36
thorstwhich can be done in relatively short order.17:36
thorstyeah, you could do that.  I don't know that everyone has done that.17:37
thorstI do that for like my github account17:37
efriedDo I have to be registered with the mailing list to post to it, or can I just post to it?17:37
efriedI'll look at the ref material...17:38
thorstyeah...not sure off hand17:39
*** smatzek has quit IRC17:44
*** smatzek has joined #openstack-powervm17:45
*** viclarson has quit IRC17:49
efriedthorst adreznec http://pastebin.com/9HkDDBNZ17:53
thorstmaybe something additional of, 'As you plan upgrades from newton to ocata, please change the nova compute driver to powervm_ext.'17:54
thorstand hop in #openstack-powervm for q's?17:54
adreznecSorry, was grabbing lunch17:58
efriedthorst: http://pastebin.com/247gdQCx17:58
adreznecThat seems reasonable.17:59
adreznecWhat tags are you planning to put on the subject?17:59
thorstand monitor for responses17:59
efried[openstack] [PowerVM OOT Driver] FYI: PowerVM out-of-tree driver namespace changing18:00
adreznecI might just do [openstack][nova][powervm] for tags18:01
efriedtoo late ;-)18:01
adreznecHeh, ok18:02
efried...unless it kicks back because my personal account isn't subscribed.18:02
efriedguess we'll see.18:02
efriedyup, rejected.18:02
*** AlexeyAbashkin has joined #openstack-powervm18:09
*** tlian has joined #openstack-powervm19:29
*** kairo has joined #openstack-powervm19:32
*** kairo has joined #openstack-powervm19:32
*** AlexeyAbashkin has quit IRC19:46
*** AlexeyAbashkin has joined #openstack-powervm20:05
*** efried has quit IRC20:12
*** kriskend_ has quit IRC20:14
*** efried has joined #openstack-powervm20:26
*** AlexeyAbashkin has quit IRC20:27
thorstefried: no luck with e-mail yet/20:35
efriedthorst, I sent it.20:35
efriedI didn't know if it would be normal for me to get my own email back from the list - but I haven't.20:35
thorstI haven't gotten it yet... adreznec did you?20:36
efriedAnd my work account is set up to get digests, so I wasn't expecting to see it for a while.20:36
adreznecI didn't see anything... let me double check20:36
efriedI hadn't sent it yet.20:37
efriedNOW I've sent it.20:37
thorstprobably takes a while to push out...20:41
*** AlexeyAbashkin has joined #openstack-powervm20:50
*** apearson has quit IRC21:00
efriedGuess maybe I was supposed to omit [openstack].  Or maybe put spaces between [tag]s?21:02
*** apearson has joined #openstack-powervm21:02
*** smatzek has quit IRC21:24
*** AlexeyAbashkin has quit IRC21:31
*** apearson has quit IRC21:31
*** kairo has quit IRC21:37
*** apearson has joined #openstack-powervm21:38
*** AlexeyAbashkin has joined #openstack-powervm21:41
*** AlexeyAbashkin has quit IRC21:47
*** svenkat has quit IRC21:49
*** antonym has quit IRC21:58
*** kriskend has joined #openstack-powervm22:11
efriedthorst, adreznec: https://review.openstack.org/#/c/391288/ tested and working.  Ready for a real review, I think.22:18
adreznecefried: Awesome, will check it out in a bit22:25
*** edmondsw has quit IRC22:27
*** edmondsw has joined #openstack-powervm22:27
*** edmondsw has quit IRC22:37
*** thorst has quit IRC22:45
*** tblakes has quit IRC22:48
*** apearson has quit IRC22:52
*** efried has quit IRC23:01
*** efried has joined #openstack-powervm23:11
openstackgerritAdam Reznechek proposed openstack/nova-powervm: Update to match latest global-requirements  https://review.openstack.org/41861623:23
openstackgerritAdam Reznechek proposed openstack/networking-powervm: Update to match latest global-requirements  https://review.openstack.org/41861723:25
openstackgerritAdam Reznechek proposed openstack/ceilometer-powervm: Update to match latest global-requirements  https://review.openstack.org/41861823:26
*** thorst has joined #openstack-powervm23:31
*** thorst has quit IRC23:35
*** kriskend has quit IRC23:42
*** kriskend has joined #openstack-powervm23:43
*** svenkat has joined #openstack-powervm23:49
*** efried has quit IRC23:58

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!