| *** thorst has joined #openstack-powervm | 00:50 | |
| *** thorst has quit IRC | 00:59 | |
| *** esberglu has joined #openstack-powervm | 01:02 | |
| *** esberglu has quit IRC | 01:06 | |
| *** thorst has joined #openstack-powervm | 02:04 | |
| *** thorst has quit IRC | 02:18 | |
| openstackgerrit | Sridhar Venkat proposed openstack/networking-powervm: WIP: Mechanism driver for powervm SR-IOV interface https://review.openstack.org/343423 | 02:30 |
|---|---|---|
| *** jwcroppe has joined #openstack-powervm | 02:48 | |
| *** thorst has joined #openstack-powervm | 03:19 | |
| *** thorst has quit IRC | 03:27 | |
| *** thorst has joined #openstack-powervm | 04:27 | |
| *** thorst has quit IRC | 04:33 | |
| *** seroyer has joined #openstack-powervm | 05:28 | |
| *** thorst has joined #openstack-powervm | 05:29 | |
| *** kotra03 has joined #openstack-powervm | 05:32 | |
| *** seroyer has quit IRC | 05:33 | |
| *** jwcroppe has quit IRC | 05:33 | |
| *** jwcroppe has joined #openstack-powervm | 05:34 | |
| *** thorst has quit IRC | 05:37 | |
| *** thorst has joined #openstack-powervm | 07:20 | |
| *** thorst has quit IRC | 07:29 | |
| *** k0da has joined #openstack-powervm | 08:09 | |
| *** thorst has joined #openstack-powervm | 08:26 | |
| *** thorst has quit IRC | 08:33 | |
| *** thorst has joined #openstack-powervm | 09:31 | |
| *** thorst has quit IRC | 09:38 | |
| *** thorst has joined #openstack-powervm | 10:36 | |
| *** thorst has quit IRC | 10:44 | |
| *** Ashana has joined #openstack-powervm | 11:38 | |
| *** edmondsw has joined #openstack-powervm | 11:40 | |
| *** erlarese has joined #openstack-powervm | 11:44 | |
| *** smatzek has joined #openstack-powervm | 11:58 | |
| *** thorst has joined #openstack-powervm | 12:05 | |
| *** k0da has quit IRC | 12:33 | |
| *** k0da has joined #openstack-powervm | 13:04 | |
| *** tblakeslee has joined #openstack-powervm | 13:07 | |
| *** mdrabe has joined #openstack-powervm | 13:09 | |
| *** jwcroppe_ has joined #openstack-powervm | 13:11 | |
| *** jwcroppe has quit IRC | 13:13 | |
| *** jwcropp__ has joined #openstack-powervm | 13:13 | |
| *** kotra03 has quit IRC | 13:15 | |
| *** jwcroppe_ has quit IRC | 13:16 | |
| *** seroyer has joined #openstack-powervm | 13:32 | |
| openstackgerrit | Sridhar Venkat proposed openstack/networking-powervm: Checkin blueprint for networking-powervm for SR-IOV VIFs support https://review.openstack.org/322210 | 13:34 |
| *** efried has quit IRC | 13:36 | |
| *** efried has joined #openstack-powervm | 13:36 | |
| *** tblakeslee has quit IRC | 14:01 | |
| *** k0da has quit IRC | 14:02 | |
| *** tpeoples has quit IRC | 14:04 | |
| *** tsjakobs has joined #openstack-powervm | 14:11 | |
| *** k0da has joined #openstack-powervm | 14:14 | |
| *** burgerk has joined #openstack-powervm | 14:15 | |
| *** tpeoples has joined #openstack-powervm | 14:18 | |
| *** tblakeslee has joined #openstack-powervm | 14:23 | |
| *** erlarese_ has joined #openstack-powervm | 14:25 | |
| *** erlarese has quit IRC | 14:28 | |
| *** erlarese_ is now known as erlarese | 14:28 | |
| *** mdrabe has quit IRC | 14:36 | |
| *** lmtaylor has joined #openstack-powervm | 14:41 | |
| *** mdrabe has joined #openstack-powervm | 14:42 | |
| *** seroyer has quit IRC | 14:51 | |
| *** esberglu has joined #openstack-powervm | 14:58 | |
| *** kriskend has joined #openstack-powervm | 15:02 | |
| *** svenkat has joined #openstack-powervm | 15:02 | |
| *** kriskend_ has joined #openstack-powervm | 15:03 | |
| svenkat | efried, thorst : this is about pci_device_id related monkey patches in sriov vif patchset : https://review.openstack.org/#/c/343419/5/nova_powervm/virt/powervm/vif.py | 15:04 |
| efried | Yuh. | 15:05 |
| efried | svenkat, can you explain how this came about in the first place? | 15:05 |
| efried | Help me with the background. | 15:05 |
| *** kotra03 has joined #openstack-powervm | 15:05 | |
| svenkat | sure, if i create a neutron port of vnic type direct and use it to deploy a VM | 15:05 |
| svenkat | there are three areas in compute code that it does not let spawn and allocate_network_instance to go through | 15:05 |
| *** seroyer has joined #openstack-powervm | 15:06 | |
| svenkat | all those three are mentioned in the patchset and monkey patched | 15:06 |
| svenkat | they are all related to pci_device_id not being present | 15:06 |
| svenkat | couple of them are in nova.network.neutronv2.api.API and one in nova.compute.claims.Claim | 15:07 |
| efried | Let's start with monkey_populate_neutron_extension_values | 15:07 |
| efried | I don't see a difference between your patched code and the original code. | 15:07 |
| efried | Can you point it out? | 15:07 |
| svenkat | sure.. one moment | 15:08 |
| svenkat | line 76, pci_request_id is set to None | 15:08 |
| svenkat | this is the difference | 15:08 |
| efried | I see. | 15:08 |
| efried | So the original code makes _populate_neutron_binding_profile a no-op; but you've activated it. | 15:10 |
| efried | sorry, vice versa. | 15:10 |
| svenkat | ok… | 15:10 |
| svenkat | yes. | 15:10 |
| efried | okay, this one should be simple enough. | 15:11 |
| svenkat | ok… | 15:11 |
| svenkat | the code is in network api in nova… is there anything in nova-powervm to overwrite it? | 15:12 |
| thorst | svenkat: or how to we change the PowerVM code to make it work with the existing behavior of nova. | 15:13 |
| thorst | thats been my stance all along | 15:13 |
| svenkat | ok… | 15:13 |
| svenkat | so how do we do it in this particular case | 15:14 |
| svenkat | next one is similar to this one. | 15:14 |
| efried | We may need to massage the pci whitelist to make that method no-op. | 15:14 |
| svenkat | ok… | 15:15 |
| svenkat | so far i could not figure out how this required value flows within nova compute code, (that has entry or exit points in nova-powervm code). | 15:17 |
| efried | Yup, looking. | 15:17 |
| svenkat | thanks | 15:18 |
| efried | thorst, svenkat: we may need to inject a dummy value into the pci whitelist to make this work. | 15:18 |
| efried | would that be a reasonable approach? | 15:18 |
| svenkat | ok… | 15:18 |
| thorst | efried: yes. | 15:19 |
| thorst | I was assuming our dummy values would be some form of our physical network names | 15:19 |
| efried | Let me look at the others and see if they would be resolved by similar means. | 15:19 |
| *** k0da has quit IRC | 15:20 | |
| efried | svenkat, this is weird, I can't find _populate_pci_mac_address at all. | 15:22 |
| svenkat | this is a new method added | 15:22 |
| svenkat | are you looking in mitaka or newton | 15:22 |
| svenkat | i had the same issue… my devstack environment has it | 15:23 |
| efried | I'm on the master branch of nova. | 15:24 |
| svenkat | ok… | 15:24 |
| *** apearson has joined #openstack-powervm | 15:26 | |
| *** lmtaylor has quit IRC | 15:28 | |
| *** lmtaylor1 has joined #openstack-powervm | 15:29 | |
| *** mdrabe has quit IRC | 15:32 | |
| *** mdrabe has joined #openstack-powervm | 15:33 | |
| thorst | svenkat: I've put back several review comments | 15:33 |
| svenkat | ok… | 15:33 |
| svenkat | in nova-powervm or networking-powervm or both? | 15:33 |
| thorst | networking-powervm. Should show up in your request view. | 15:34 |
| svenkat | sure.. looking | 15:34 |
| efried | thorst, how do I go from somebody's name to their IRC handle? | 15:36 |
| thorst | efried: good q...who are you looking for? | 15:36 |
| efried | Nikola Dipanov | 15:36 |
| thorst | not sure... | 15:38 |
| thorst | https://www.google.com/search?q=Nikola+Dipanov&ie=utf-8&oe=utf-8 | 15:38 |
| thorst | probably djipko | 15:39 |
| efried | Yeah, that's what I was seeing. | 15:39 |
| efried | Can't see him in IRC - at least not the channels I'm searchin.g | 15:39 |
| thorst | works for Facebook...still a little early for west coast. | 15:40 |
| efried | Well, svenkat, it appears to me as though the _populate_pci_mac_address method was removed at some point in the last three weeks. | 15:44 |
| efried | So that issue may be moot. | 15:44 |
| svenkat | oh ok… | 15:44 |
| thorst | efried FTW | 15:44 |
| thorst | :-) | 15:45 |
| efried | It would be tough to figure out when that happened; but I can do that if it's necessary. | 15:45 |
| svenkat | thats good. if it is gone, we can ignore it | 15:45 |
| efried | But, in case you needed any further encouragement, this is a dramatic case-in-point why we don't monkey patch - especially private methods. | 15:45 |
| openstackgerrit | Sridhar Venkat proposed openstack/networking-powervm: Checkin blueprint for networking-powervm for SR-IOV VIFs support https://review.openstack.org/322210 | 15:45 |
| svenkat | ok… | 15:46 |
| thorst | efried: I'd like a +1 from you on this before I W+1 it: https://review.openstack.org/#/c/322210/6 | 15:50 |
| efried | thorst, time frame? | 15:56 |
| thorst | efried: dept lunch | 15:56 |
| efried | That doesn't help me. | 15:56 |
| *** thorst is now known as thorst_afk | 15:56 | |
| *** efried has quit IRC | 16:08 | |
| *** kotra03 has quit IRC | 16:55 | |
| *** thorst_afk is now known as thorst | 17:19 | |
| thorst | esberglu: I remapped all the volumes to neo6 | 17:19 |
| thorst | can you re-add to ssp? do you know those steps? | 17:19 |
| esberglu | thorst: You just go on the vios and do cluster addnode right? | 17:40 |
| esberglu | Which vios owns the cluster though? | 17:42 |
| esberglu | thorst: nvm. found it | 17:45 |
| *** mdrabe has quit IRC | 18:46 | |
| *** mdrabe has joined #openstack-powervm | 19:09 | |
| adreznec | thorst: You good on https://review.openstack.org/#/c/348544/ merging then? | 19:13 |
| *** efried has joined #openstack-powervm | 19:19 | |
| thorst | adreznec: looking... | 19:33 |
| thorst | adreznec: yeah, definitely | 19:35 |
| thorst | merge away | 19:35 |
| *** k0da has joined #openstack-powervm | 19:57 | |
| *** jwcropp__ has quit IRC | 20:47 | |
| *** jwcroppe has joined #openstack-powervm | 20:48 | |
| *** jwcroppe has quit IRC | 20:48 | |
| *** jwcroppe has joined #openstack-powervm | 20:48 | |
| openstackgerrit | Merged openstack/nova-powervm: Fix package setup configuration https://review.openstack.org/348544 | 20:53 |
| efried | thorst, mdrabe: brainstorming solutions for the stale mappings bug again. | 20:54 |
| thorst | k...I need to head out in 15-20 | 20:54 |
| thorst | so lets brainstorm quick | 20:54 |
| efried | What I'm thinking now is simply to add a mapping scrub at the end of the destroy flow to clean up any leftovers. | 20:54 |
| mdrabe | you'd still hit that warning | 20:55 |
| efried | Don't care. | 20:55 |
| efried | Do I? | 20:55 |
| thorst | is a warning a bad thing? | 20:55 |
| thorst | We could change the message. | 20:55 |
| thorst | to be a little friendlier | 20:55 |
| mdrabe | The warning is saying: we're not gonna clean up this mapping or hdisk | 20:55 |
| mdrabe | but then you're gonna end up cleaning it up | 20:55 |
| thorst | ew | 20:56 |
| efried | Yeah; we would need to change that. | 20:56 |
| efried | "Disconnect Volume: The backing hdisk for volume " | 20:56 |
| efried | "%(volume_id)s on Virtual I/O Server %(vios)s is " | 20:56 |
| efried | "not in a valid state. No disconnect " | 20:56 |
| efried | "actions to be taken as volume is not healthy."), | 20:56 |
| efried | Just whack that last sentence and I think we're okay. | 20:57 |
| thorst | efried: What other flows could perhaps be impacted by that? | 20:58 |
| thorst | cold migration? | 20:58 |
| thorst | (or does that call down to destroy...I think it does?) | 20:58 |
| efried | Sure, any flow that's using destroy would be impacted, but I believe that's what we want. | 20:59 |
| efried | In destroy, we always want all VSCSI mappings for the instance to go away. | 20:59 |
| *** svenkat has quit IRC | 21:00 | |
| mdrabe | If you're gonna change that warning message it might be nice to add something like "...is not in a valid state. This could be the result of evacuation" | 21:01 |
| mdrabe | Those stale mappings could be tackled from there too though | 21:02 |
| thorst | efried: what about disconnect volume...where you remove a volume but don't destroy the server? | 21:02 |
| efried | "server"? | 21:03 |
| efried | oh, the instance. | 21:03 |
| efried | Yeah, the plan would be to put the scrub into destroy, not into disconnect_volume. | 21:03 |
| thorst | efried: yep, sorry | 21:04 |
| thorst | I'm just wondering if it could leave a volume around that way that we'd clean up otherwise | 21:04 |
| thorst | nonetheless, its not like we're taking anything away from what we have today | 21:04 |
| thorst | rather, just doing additional clean up.... | 21:04 |
| efried | Right. Theoretically, disconnect_volume should be destroying any disk that we can actually find. | 21:04 |
| thorst | so I'm leaning towards liking it | 21:04 |
| efried | The problem we're having today is that, when the disk has been ripped out, we can't even get a name/UDID to use to identify the mapping. | 21:05 |
| thorst | efried: well, it'll destory a disk on the vscsi disconnect I think | 21:05 |
| thorst | yeah... | 21:05 |
| thorst | so I think that's fine | 21:05 |
| efried | So if you try to disconnect_volume in a non-destroy flow, and that disk has been ripped out, you'll still leave that stale mapping around. | 21:05 |
| efried | that's still a hole | 21:05 |
| thorst | but would be cleaned up on a destroy | 21:06 |
| efried | yes. | 21:06 |
| thorst | so not a significant hole IMO | 21:06 |
| efried | And I don't think we want to go removing storage-less mappings - because if the disk gets restored at some point, we want it to reappear to the VM. | 21:06 |
| thorst | fair enuf | 21:06 |
| efried | I.e. handle flaky storage backend. | 21:06 |
| efried | Okay, so now the only question is whether I can use VSCSIBus.delete() or have to do a second VIOS feed GET with VSCSI mapping xags. | 21:07 |
| mdrabe | shouldn't the Feed already be there? | 21:07 |
| mdrabe | With xags galore | 21:07 |
| thorst | everyone gets a xag | 21:08 |
| thorst | gotta run, but I'm currently with mdrabe...I'd be thinking that the feed is already there | 21:08 |
| *** thorst has quit IRC | 21:08 | |
| *** thorst has joined #openstack-powervm | 21:09 | |
| *** thorst has quit IRC | 21:13 | |
| efried | argh, sorry kids. mdrabe, we *may* be able to use the same feed. I'll have to look into that. | 21:15 |
| *** thorst has joined #openstack-powervm | 21:34 | |
| *** lmtaylor1 has quit IRC | 21:35 | |
| *** thorst has quit IRC | 21:38 | |
| *** burgerk has quit IRC | 21:46 | |
| *** smatzek has quit IRC | 22:03 | |
| *** jwcroppe_ has joined #openstack-powervm | 22:04 | |
| *** esberglu has quit IRC | 22:04 | |
| *** jwcroppe has quit IRC | 22:06 | |
| *** mdrabe has quit IRC | 22:08 | |
| *** mdrabe has joined #openstack-powervm | 22:09 | |
| *** mdrabe has quit IRC | 22:17 | |
| *** k0da has quit IRC | 22:19 | |
| *** Ashana has quit IRC | 22:24 | |
| *** esberglu has joined #openstack-powervm | 22:25 | |
| *** tblakeslee has quit IRC | 22:26 | |
| *** efried has quit IRC | 22:28 | |
| *** seroyer has quit IRC | 22:30 | |
| *** Ashana has joined #openstack-powervm | 22:31 | |
| *** Ashana has quit IRC | 22:35 | |
| *** kriskend_ has quit IRC | 22:40 | |
| *** kriskend has quit IRC | 22:40 | |
| *** Ashana has joined #openstack-powervm | 22:41 | |
| *** Ashana has quit IRC | 22:45 | |
| *** esberglu has quit IRC | 22:45 | |
| *** esberglu has joined #openstack-powervm | 22:49 | |
| *** thorst has joined #openstack-powervm | 22:51 | |
| *** Ashana has joined #openstack-powervm | 22:52 | |
| *** Ashana has quit IRC | 22:56 | |
| *** tsjakobs has quit IRC | 22:57 | |
| *** thorst_ has joined #openstack-powervm | 23:00 | |
| *** thorst has quit IRC | 23:02 | |
| *** Ashana has joined #openstack-powervm | 23:04 | |
| *** Ashana has quit IRC | 23:08 | |
| *** Ashana has joined #openstack-powervm | 23:12 | |
| *** thorst_ has quit IRC | 23:13 | |
| *** thorst has joined #openstack-powervm | 23:14 | |
| *** Ashana has quit IRC | 23:17 | |
| *** esberglu has quit IRC | 23:17 | |
| *** Ashana has joined #openstack-powervm | 23:18 | |
| *** thorst has quit IRC | 23:22 | |
| *** Ashana has quit IRC | 23:22 | |
| *** esberglu has joined #openstack-powervm | 23:25 | |
| *** tblakeslee has joined #openstack-powervm | 23:26 | |
| *** Ashana has joined #openstack-powervm | 23:30 | |
| *** Ashana has quit IRC | 23:34 | |
| *** tsjakobs has joined #openstack-powervm | 23:35 | |
| *** Ashana has joined #openstack-powervm | 23:35 | |
| *** edmondsw has quit IRC | 23:40 | |
| *** Ashana has quit IRC | 23:40 | |
| *** tsjakobs has quit IRC | 23:40 | |
| *** Ashana has joined #openstack-powervm | 23:41 | |
| *** Ashana has quit IRC | 23:46 | |
| *** Ashana has joined #openstack-powervm | 23:47 | |
| *** Ashana has quit IRC | 23:51 | |
| *** Ashana has joined #openstack-powervm | 23:53 | |
| *** thorst has joined #openstack-powervm | 23:54 | |
| *** jwcroppe has joined #openstack-powervm | 23:56 | |
| *** Ashana has quit IRC | 23:57 | |
| *** jwcroppe_ has quit IRC | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!