Thursday, 2017-12-07

*** edmondsw has quit IRC01:12
*** edmondsw has joined #openstack-powervm02:33
*** edmondsw has quit IRC02:37
*** chhavi has joined #openstack-powervm04:37
*** AlexeyAbashkin has joined #openstack-powervm06:15
*** AlexeyAbashkin has quit IRC06:26
*** AlexeyAbashkin has joined #openstack-powervm08:07
catmandocsky_:  it turns out to be very weird. the replacement of the switches was done correctly, and the fabrics were reconfigured correctly. however, even after a restart, powervc was not seeing the fabric because the ssh keys had changed. what you have to do is go the the fabric configuration and log in again to each switch in the fabric (and accept the ssh key change). it then seems to take about 15 minutes11:37
catmandofor the fabric to be recognised again11:37
*** csky_ has quit IRC12:13
*** apearson has joined #openstack-powervm13:29
*** svenkat has joined #openstack-powervm13:31
*** AlexeyAbashkin has quit IRC14:12
*** chhavi has quit IRC14:12
*** chhavi has joined #openstack-powervm14:13
*** esberglu has joined #openstack-powervm14:16
*** chhavi__ has joined #openstack-powervm14:17
*** chhavi has quit IRC14:20
*** edmondsw has joined #openstack-powervm14:30
*** toan has quit IRC14:48
*** ChanServ has quit IRC14:48
*** chhavi__ has quit IRC14:48
*** efried has quit IRC14:48
*** catmando has quit IRC14:48
*** mdrabe has quit IRC14:48
*** svenkat has quit IRC14:48
*** tonyb has quit IRC14:48
*** toan has joined #openstack-powervm14:54
*** chhavi__ has joined #openstack-powervm14:54
*** svenkat has joined #openstack-powervm14:54
*** efried has joined #openstack-powervm14:54
*** catmando has joined #openstack-powervm14:54
*** mdrabe has joined #openstack-powervm14:54
*** tonyb has joined #openstack-powervm14:54
*** chhavi__ has quit IRC14:56
esbergluedmondsw: Talked with efried yesterday about the vscsi change and we think the nova guys will be more likely to be onboard with a single collapsed class15:14
esbergluYou guys want to discuss further?15:14
*** ChanServ has joined #openstack-powervm15:16
*** sets mode: +o ChanServ15:16
edmondswefried why do you think they'll care?15:25
efriedlemme pull up my rationale from yesterday..15:26
edmondswif the breakdown makes sense, and they know we're going to be expanding on this...15:26
efriededmondsw Yeah, problem is a) the breakdown doesn't make a lot of sense here, especially in isolation; and b) we're not going to be expanding until Rocky.15:27
edmondswI think (a) is the real key there, if true... but I thought the breakdown actually made sense (with the redesign we'd discussed)15:27
efriedI think we could get away with it if it was just one superclass with abstract methods and default impls.15:28
efriedI'm pretty skeptical about the whole base.connect_volume calling sub._connect_volume, though.  That pattern is going to be hard to justify with, "but it's important later".15:29
efriedI mean, it is (important later), but that's going over the line of ugly for a patch that's going to sit alone in a release.15:29
edmondswefried I'm ok with one base class and one impl for now15:31
edmondswisn't base.connect_volume calling sub._connect_volume a pretty common/standard thing?15:31
edmondswwhenever you've got something where you want a partially common and partially specific impl, how else would you do it?15:31
efriedIs it?  I cringe whenever I see it.  I would hope it's only done when absolutely necessary.  And again, in a single impl patch, it ain't.15:32
edmondswok, compress it down...15:32
edmondswesberglu ^15:32
efriedSo I would still prefer folded into a single class; but would not oppose a base + impl as long as it doesn't do THAT thing.15:33
edmondswmaybe when we split it later someone will have a better suggestion than _connect_volume, which they couldn't have now missing the necessary context15:33
catmandoedmondsw: thanks for the help yesterday, it turned out to be a strange bit of powervc behavior15:37
esbergluedmondsw: efried: Okay. I'll get going on that today15:38
esbergluedmondsw: Btw I stood up an IT SEA cloud yesterday15:38
esbergluThings are looking good, the vifs are being created, you can ssh in, the instance can reach the outside15:39
edmondswcatmando np. I hope you got it figured out?15:39
edmondswesberglu excellent15:39
catmandoyes, it turns out that you have to re-login to the fabric manually to reset ssh keys15:39
edmondswesberglu any more luck with OVS?15:39
catmandothere does not seem to be any api call for it15:40
esbergluInstalled a diff. machine for OVS, got a cloud stood up (OOT), spawns aren't working yet though15:40
edmondswesberglu at least it sounds like you're making progress15:41
edmondswcatmando I'll run that by csky and see what he thinks15:42
edmondswcatmando I'm not sure I followed you15:42
edmondswor rather, I'm sure I didn't quite15:42
catmandochanging the switches also changed their ssh signatures, and powervc does not log-in to the fabric (as set at SOMEIP/powervc/#t=storageVolumes.fabric_details&selectedTab=fabrics&fabric_name=SOMENAME) upon reboot15:44
catmandoyou have to go to that page and manually re-enter the password. at that point powervc will warn you about the new signature and once confirmed something happens in the background that eventually (about 15 minutes later) makes the fabric available again within the SCG15:45
edmondswcatmando oh, you had to edit the registration of the switches in PowerVC?15:48
edmondswthat I can understand15:49
edmondswa little disappointed if it didn't give you a health warning about the switches before you did that, though. At least if you had run "verify environment" it should have warned you that we couldn't talk to the fabric switches, but I'm guessing you didn't do that. I would have hoped it would warn you anyway the first time we tried to talk to the switch and couldn't.15:51
catmando@edmondsw apologies if this was an annoying one, i'm still learning. the environment gave no warning that something went wrong, and the error is not documented anywhere. once you guys explained it, it makes sense15:58
edmondswcatmando sorry we didn't make it more obvious :)15:58
edmondswwe normally pride ourselves on ease of use15:58
catmandoi am very thankful for the help and patience15:58
catmandoi know that this is not directly powervc related16:24
catmandobut any thoughts on where I may get the actual documentation for Spectrum Control's REST API?16:24
catmando  <----------- this doesn't actually seem to give me a list of endpoints, just a few examples16:24
*** AlexeyAbashkin has joined #openstack-powervm16:24
efriedthorst would be the one to answer that.16:25
efriedNever mind - he knows spectrum *scale*, which is apparently different.16:29
catmandoif scale can create and edit volume properties on a storwize...16:30
catmandoi don't know it, but if it's restful i would be willing to try16:30
catmandoanything is better than in-code ssh16:31
*** AlexeyAbashkin has quit IRC16:31
catmandoyeah that documentation looks complete16:32
catmandomany thanks, shall investigate16:32
*** apearson has quit IRC16:38
*** esberglu has quit IRC18:24
*** esberglu has joined #openstack-powervm18:25
*** esberglu_ has joined #openstack-powervm18:29
*** esberglu has quit IRC18:30
*** edmondsw has quit IRC19:00
*** esberglu_ is now known as esberglu19:33
*** edmondsw has joined #openstack-powervm19:35
*** edmondsw has quit IRC19:39
*** apearson has joined #openstack-powervm19:40
*** edmondsw has joined #openstack-powervm19:41
*** edmondsw has quit IRC19:45
*** edmondsw has joined #openstack-powervm19:52
*** edmondsw has quit IRC19:56
*** edmondsw has joined #openstack-powervm19:58
*** edmondsw has quit IRC19:58
*** edmondsw has joined #openstack-powervm19:58
*** edmondsw has quit IRC20:02
*** edmondsw has joined #openstack-powervm20:07
*** edmondsw has quit IRC20:12
*** esberglu has quit IRC21:43
*** esberglu has joined #openstack-powervm21:48
esbergluefried: I should be renaming all of the connect/disconnects as attach/detach right?21:49
efriedShoot, I don't know - was that something edmondsw brought up?21:50
esbergluWe renamed ConnectDisk and DisconnectDisk to attach/detach previously, I think the nova guys requested that?21:51
esbergluI have a decent start on the vSCSI stuff here
esbergluStill plenty to do, I need to cleanup quite a bit, but wouldn't mind an initial review before you're out21:52
*** svenkat has quit IRC22:03
*** apearson has quit IRC22:12
*** esberglu has quit IRC22:36
*** esberglu has joined #openstack-powervm22:42
*** esberglu has quit IRC22:50

Generated by 2.15.3 by Marius Gedminas - find it at!