| *** openstack has joined #openstack-powervm | 02:35 | |
| *** ChanServ sets mode: +o openstack | 02:35 | |
| *** esberglu has quit IRC | 13:44 | |
| *** esberglu has joined #openstack-powervm | 13:45 | |
| *** esberglu has quit IRC | 13:50 | |
| *** esberglu has joined #openstack-powervm | 13:56 | |
| openstackgerrit | Matthew Edmonds proposed openstack/ceilometer-powervm master: docs: Modernize build process https://review.openstack.org/592057 | 14:08 |
|---|---|---|
| efried | edmondsw: Does pvc allow hot-plugging SR-IOV VFs to active VMs? | 14:13 |
| edmondsw | I believe so, but I'll confirm | 14:14 |
| edmondsw | efried yes, confirmed | 14:15 |
| efried | edmondsw: Do we do it via the attach_interface API? If so, then we'll need to hold a local revert of https://review.openstack.org/#/c/591898/ when it lands :( | 14:16 |
| efried | and/or possibly address the allocation problem some other way? | 14:16 |
| edmondsw | efried yes it would be using attach_interface | 14:18 |
| edmondsw | efried I'm trying to determine if this will also break nova-powervm/networking-powervm. I suspect yes | 14:27 |
| edmondsw | if so, I think we can make a good case that that commit is not hypervisor agnostic and should not merge | 14:27 |
| efried | If there's an alloc accounting snafu, we may get steamrolled | 14:28 |
| efried | but yeah, would be worth a shot. | 14:28 |
| efried | cause off the cuff, I don't think they're denying it takes away capability. | 14:29 |
| edmondsw | I think we're reading it differently | 14:29 |
| edmondsw | I thought mriedem was implying it would fail anyway, and he just wanted it to fail fast | 14:30 |
| edmondsw | let me go look at the bug... | 14:30 |
| edmondsw | efried talking to erlarese and yes, it sounds like it would break nova-powervm/networking-powervm's SRIOV attach capability | 14:33 |
| efried | woot | 14:33 |
| efried | The push back is going to look like one or more of: | 14:34 |
| efried | - Your support is OOT and we don't care about OOT | 14:34 |
| efried | - The spec says we don't support it, so we shouldn't support it | 14:34 |
| edmondsw | we support SEA and OVS with the IT driver... is there more work to support SRIOV there? | 14:34 |
| efried | um | 14:35 |
| efried | I think so | 14:35 |
| edmondsw | I'm guessing yes... ok | 14:35 |
| efried | We have a PvmVnicSriovVifDriver in nova-powervm, so yeah. | 14:36 |
| efried | We may be able to argue back that | 14:36 |
| efried | - specs aren't support statements | 14:36 |
| efried | - you can't take away working function without a microversion | 14:36 |
| edmondsw | yep | 14:36 |
| efried | And maybe suggest that we should instead make a support matrix block for sriov hot-plug and put 'missing' everywhere | 14:37 |
| efried | instead of doing this fast-fail | 14:37 |
| efried | because then we (nova-powervm) could claim support (at least partial) in our support matrix. | 14:37 |
| efried | So | 14:37 |
| efried | Do you want to do this -1? | 14:37 |
| edmondsw | yes, I'm working on it | 14:37 |
| edmondsw | tx for raising it | 14:37 |
| efried | I think it would play best from you as the nova-powervm ptl. And I can jump in and support you as needed. | 14:38 |
| edmondsw | yep | 14:38 |
| edmondsw | efried comment is up | 14:43 |
| edmondsw | tried to keep it short and sweet | 14:43 |
| efried | nice, I like it. | 14:44 |
| *** gman-tx has joined #openstack-powervm | 16:16 | |
| gman-tx | efried are you ready for me to +2 this? https://review.openstack.org/#/c/468560/ | 16:20 |
| efried | gman-tx: You could try. I would be happy with a +1 though :) (It's nova) | 16:20 |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm master: PowerVM: update_provider_tree() (compatible) https://review.openstack.org/591857 | 16:22 |
| gman-tx | i can only +1 in nova ... | 16:22 |
| efried | gman-tx: You could +2 this one https://review.openstack.org/#/c/591857/ (<== edmondsw fixed the import order) | 16:22 |
| gman-tx | efried will have a look | 16:24 |
| efried | gman-tx: Thx. It's the same patch, just in nova-powervm. | 16:24 |
| gman-tx | efried deja vu | 16:26 |
| efried | thx gman-tx | 16:28 |
| *** efried is now known as efried_rollin | 16:28 | |
| gman-tx | welcome efried_rollin | 16:29 |
| *** openstackstatus has joined #openstack-powervm | 16:57 | |
| *** ChanServ sets mode: +v openstackstatus | 16:57 | |
| openstackgerrit | Merged openstack/nova-powervm master: PowerVM: update_provider_tree() (compatible) https://review.openstack.org/591857 | 17:26 |
| *** gman-tx has quit IRC | 17:31 | |
| *** openstackgerrit has quit IRC | 19:19 | |
| *** efried_rollin is now known as efried | 19:53 | |
| *** openstackgerrit has joined #openstack-powervm | 20:07 | |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm master: Inventory whitelist schema and loading https://review.openstack.org/579289 | 20:07 |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm master: Filter IOSlots per inventory config yaml https://review.openstack.org/589306 | 20:07 |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm master: Placement-isms for inventory items https://review.openstack.org/589307 | 20:07 |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm master: WIP: update_provider_tree with device exposure https://review.openstack.org/589668 | 20:07 |
| efried | *really* don't understand why this caused a merge conflict, but whatever. | 20:07 |
| *** esberglu has quit IRC | 22:02 | |
| *** esberglu has joined #openstack-powervm | 23:15 | |
| *** esberglu has quit IRC | 23:20 | |
| *** esberglu has joined #openstack-powervm | 23:40 | |
| *** esberglu has quit IRC | 23:45 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!