Friday, 2017-08-25

*** esberglu has joined #openstack-powervm00:32
*** apearson has joined #openstack-powervm01:32
*** esberglu has quit IRC01:42
*** edmondsw has quit IRC02:55
*** apearson has quit IRC02:59
*** edmondsw has joined #openstack-powervm04:40
*** edmondsw has quit IRC04:45
*** tjakobs has joined #openstack-powervm06:10
*** edmondsw has joined #openstack-powervm06:29
*** tjakobs has quit IRC06:30
*** edmondsw has quit IRC06:33
*** k0da has joined #openstack-powervm08:06
*** chas has joined #openstack-powervm08:11
*** edmondsw has joined #openstack-powervm08:17
*** edmondsw has quit IRC08:21
*** k0da has quit IRC08:22
*** k0da has joined #openstack-powervm08:41
*** k0da has quit IRC08:59
*** openstackgerrit has quit IRC10:02
*** edmondsw has joined #openstack-powervm10:05
*** edmondsw has quit IRC10:09
*** k0da has joined #openstack-powervm10:10
*** chas_ has joined #openstack-powervm10:17
*** chas has quit IRC10:20
*** k0da has quit IRC11:03
*** smatzek has joined #openstack-powervm11:19
*** k0da has joined #openstack-powervm11:26
*** mdrabe has quit IRC11:41
*** toan has quit IRC11:41
*** chas_ has quit IRC11:41
*** miltonm has quit IRC11:41
*** ChanServ has quit IRC11:41
*** kylek3h has quit IRC11:41
*** AndyWojo has quit IRC11:41
*** k0da has quit IRC11:41
*** smatzek has quit IRC11:41
*** thorst_afk has quit IRC11:41
*** efried has quit IRC11:41
*** adi_____ has quit IRC11:41
*** zerick has quit IRC11:41
*** adreznec has quit IRC11:41
*** tonyb has quit IRC11:41
*** openstack has joined #openstack-powervm12:58
*** apearson has joined #openstack-powervm12:58
*** openstackstatus has joined #openstack-powervm12:58
*** ChanServ sets mode: +v openstackstatus12:58
*** tjakobs has joined #openstack-powervm13:28
*** esberglu has joined #openstack-powervm13:37
edmondswefried it seems the biggest problem with using these fake PCI addresses is that an operator filling out the conf won't know how to get that...13:47
edmondswwhy does nova care about the format of those addresses?13:47
efriedThe operator doesn't need to know about the fake PCI addresses as yet.13:47
edmondswthey will to fill out a conf that adds address= to the whitelist13:48
efriedbecause they assume they can access /sys/bus/.../<pci_address>/blah/blah13:48
efriedYeah, so if we need to do that, and we get to a point where we *can* do that, we're going to need to provide access to the spoofed PCI addresses via pvmctl.13:48
edmondswugh13:48
efriedWe're going to need to doc how to get the vendor & product ID from pvmctl anyway.13:49
efriedCause they're called pci_vendor_id and pci_dev_id, respectively, right now.13:49
efriedWe could add @property aliases for them to allow them to be gotten by a more expected name.13:49
edmondswsure, but that makes sense, vendor/product IDs are real things, and they're already there13:50
efriedAnd we could add a @property for pci_address too - though I would want to name it carefully so as not to imply that it's a *real* PCI address.13:50
efriedCould do it via pvmctl, or via some other CLI utility we ship with nova-powervm.13:50
edmondswI get that nova assumes they can access /sys/bus... but they don't need to here, so they shouldn't care in this flow... i.e. we should be able to propose nova changes to no longer care about this format, right?13:50
efriedI'm still on the fence about whether to do the PCI address spoofing in pypowervm or in nova.13:51
edmondswI'd rather go down the "fix nova" road until we hit a roadblock there, before going down the "hack pvm" road13:51
efriedYou mean so we can use something arbitrary (like the DRC index) as the address?13:51
edmondswyeah, whatever makes sense... I thought there was something more user friendly than drc13:52
efriedThe location code.  But same point.13:52
edmondswyep13:52
efriedRight now the whitelist code is heavily invested in storing and validating PCI addresses as domain:bus:slot.func.13:53
efriedAnd doing the whitelist x alias filtering based on that format13:53
efriedincluding the ability to wildcard any of those components.13:53
edmondswright... that rings a bell... so that's what would have to change13:53
efriedSo if you whitelisted *:12:ab.*, you would match a device with address abcd:12:ab.7 but not abcd:13:ab.713:54
efriedYeah, I would expect a LOT of resistance getting that setup overhauled.13:54
efriedThat said, it *may* be the case that the new resource provider work would lend itself to bypassing that entire shenanigans.13:56
edmondswwe can try, at least :) I thought talking to Jay in Boston he agreed that this was a mess and really shouldn't be PCI address-specific13:57
edmondswbut of course that's just one person we'd need to win over13:57
efriededmondsw I'm talking about it in -nova.  Don't yet have the attention of the right people, but stephenfin and cdent are listening.14:07
edmondswefried cool... I'm on a call, but I'll try to keep up as I can14:08
*** tjakobs has quit IRC14:19
*** smatzek has quit IRC14:44
*** efried is now known as fried_rice14:46
*** mdrabe_ has joined #openstack-powervm14:58
*** mdrabe has quit IRC15:01
*** mdrabe has joined #openstack-powervm15:03
*** mdrabe_ has quit IRC15:07
*** dwayne_ has joined #openstack-powervm15:12
*** k0da has quit IRC15:21
*** apearson_ has joined #openstack-powervm15:36
*** miltonm_ has joined #openstack-powervm15:36
*** dwayne__ has joined #openstack-powervm15:36
*** tjakobs has joined #openstack-powervm15:36
*** dwayne_ has quit IRC15:39
*** apearson has quit IRC15:39
*** miltonm has quit IRC15:39
*** miltonm__ has joined #openstack-powervm15:40
*** dwayne_ has joined #openstack-powervm15:41
*** apearson__ has joined #openstack-powervm15:41
*** miltonm_ has quit IRC15:44
*** apearson_ has quit IRC15:44
*** dwayne__ has quit IRC15:44
*** openstackgerrit has joined #openstack-powervm16:34
openstackgerritMatt Rabe proposed openstack/nova-powervm master: Add PPT ratio value to supported flavor fields  https://review.openstack.org/49797716:34
*** esberglu has quit IRC16:43
*** esberglu has joined #openstack-powervm16:43
*** esberglu has quit IRC16:44
*** esberglu has joined #openstack-powervm17:11
*** tjakobs_ has joined #openstack-powervm17:59
*** tjakobs has quit IRC17:59
openstackgerritTaylor Jakobson proposed openstack/nova-powervm master: Add support for ceph volumes  https://review.openstack.org/48447818:18
*** chas_ has quit IRC18:55
*** apearson__ has quit IRC20:08
*** miltonm__ has quit IRC20:08
*** apearson__ has joined #openstack-powervm20:11
*** dwayne_ has quit IRC20:23
*** esberglu has quit IRC20:29
*** thorst_afk has quit IRC20:53
*** chas has joined #openstack-powervm20:55
*** chas has quit IRC21:00
openstackgerritTaylor Jakobson proposed openstack/nova-powervm master: Add support for ceph volumes  https://review.openstack.org/48447821:04
*** thorst_afk has joined #openstack-powervm21:10
*** thorst_afk has quit IRC21:15
*** edmondsw has quit IRC21:25
*** thorst_afk has joined #openstack-powervm22:11
*** apearson__ has quit IRC22:12
*** thorst_afk has quit IRC22:15
*** thorst_afk has joined #openstack-powervm22:27
*** thorst_afk has quit IRC22:31
*** thorst_afk has joined #openstack-powervm22:31
*** miltonm has joined #openstack-powervm22:35
*** thorst_afk has quit IRC22:36
*** tjakobs_ has quit IRC22:37
*** fried_rice is now known as efried22:46
*** chas has joined #openstack-powervm22:57
*** chas has quit IRC23:01
*** miltonm has quit IRC23:19

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!