*** edmondsw has joined #openstack-powervm | 01:04 | |
*** edmondsw has quit IRC | 01:08 | |
*** edmondsw has joined #openstack-powervm | 02:52 | |
*** edmondsw has quit IRC | 02:57 | |
*** edmondsw has joined #openstack-powervm | 04:40 | |
*** edmondsw has quit IRC | 04:44 | |
*** edmondsw has joined #openstack-powervm | 06:28 | |
*** edmondsw has quit IRC | 06:33 | |
openstackgerrit | Chhavi Agarwal proposed openstack/nova-powervm master: iSCSI volume detach with no UDID https://review.openstack.org/576034 | 07:27 |
---|---|---|
*** edmondsw has joined #openstack-powervm | 08:16 | |
*** edmondsw has quit IRC | 08:21 | |
*** edmondsw has joined #openstack-powervm | 12:24 | |
*** edmondsw has quit IRC | 12:29 | |
*** edmondsw has joined #openstack-powervm | 12:31 | |
*** edmondsw has quit IRC | 12:35 | |
*** edmondsw has joined #openstack-powervm | 12:37 | |
*** edmondsw has quit IRC | 12:42 | |
*** edmondsw has joined #openstack-powervm | 12:45 | |
*** efried_off is now known as efried | 12:46 | |
*** efried has quit IRC | 12:46 | |
*** edmondsw has quit IRC | 12:49 | |
*** edmondsw has joined #openstack-powervm | 12:51 | |
*** edmondsw has quit IRC | 12:53 | |
*** edmondsw has joined #openstack-powervm | 12:53 | |
*** efried has joined #openstack-powervm | 13:02 | |
openstackgerrit | Merged openstack/nova-powervm master: Fix requirements and lower constraints https://review.openstack.org/579701 | 16:04 |
openstackgerrit | Merged openstack/nova-powervm master: Bump pypowervm min 1.1.16 https://review.openstack.org/579248 | 16:05 |
openstackgerrit | Merged openstack/nova-powervm master: Add support for rbd user https://review.openstack.org/573837 | 16:06 |
edmondsw | efried when you get a chance: https://review.openstack.org/#/c/580493/ | 16:55 |
*** chhagarw has joined #openstack-powervm | 16:58 | |
chhagarw | edmondsw: can we close on this review: https://review.openstack.org/#/c/576034/ | 16:58 |
chhagarw | I updated the comments | 16:59 |
edmondsw | chhagarw looking | 17:03 |
edmondsw | chhagarw have you asked Gerald to review? | 17:04 |
chhagarw | yeah, i reminded him as well | 17:04 |
edmondsw | chhagarw posted comments | 17:12 |
chhagarw | so u mean we need to remove this with block completely https://review.openstack.org/#/c/576034/11/nova_powervm/tests/virt/powervm/volume/test_iscsi.py@395 | 17:18 |
chhagarw | no need to check mock_discover | 17:18 |
edmondsw | yes | 17:19 |
efried | edmondsw: I've got that one queued up. I want to pull it down and try it, but I wanted to clear out my sandbox first. | 17:32 |
edmondsw | efried tx | 17:33 |
efried | edmondsw: Talk to me about support for attaching GPUs to existing instances. That's a use case we want to support, nicht wahr? | 17:34 |
edmondsw | efried correct | 17:37 |
efried | edmondsw: To my knowledge, there's no ComputeDriver interface for "attach PCI device". Do you know of one? | 17:38 |
edmondsw | no, don't know of one | 17:38 |
edmondsw | and I would not get too hung up on that for the first go around | 17:38 |
edmondsw | I'm not sure if that can be done hot or would require shutdown | 17:39 |
edmondsw | and I don't think it will be super common... more an "eventually I expect this will be wanted" thing | 17:39 |
efried | edmondsw: In the general case it can certainly be done hot as far as PowerVM is concerned, hot-attaching a slot to a VM. Don't know about GPUs specifically or what it would take to discover and make usable in the guest. | 17:40 |
efried | But hot or cold, it would be the same operation from a ComputeDriver interface point of view. | 17:40 |
edmondsw | right, powervm can definitely hot-attach PCI... but some individual cards support that and some don't, and I don't know which is the case for current GPUs | 17:41 |
efried | Unless now we'd be saying the only support would be via resize. | 17:41 |
efried | But okay, scope of current work is spawn/destroy only. Rolling with that... | 17:41 |
edmondsw | I had thought it would be via resize because that's just how nova works today, though it is odd that pci is handled differently from storage and networks in that way | 17:42 |
efried | I guess they care more about being able to hot-plug those things. | 17:42 |
edmondsw | and yeah, I guess we have to implement something either to make it work or block it, since someone could request more or less PCI via resize with current interface | 17:43 |
efried | we should get it for free via resize, nah? | 17:44 |
edmondsw | that'd sure be nice | 17:46 |
edmondsw | I don't know, honestly. Will phyp block a hot attach for cards that don't support that, in which case we have to handle some kind of failure? | 17:47 |
edmondsw | or... I don't know | 17:47 |
efried | No, not hot attach. Resize only applies to cold migrate. Instance shut down. | 17:48 |
efried | You can't specify a new flavor for live migrate. | 17:48 |
efried | Though it raises the question about whether we ought to block or support live migrate with attached GPUs. | 17:49 |
efried | Gotta say no, because we can't migrate the state of the GPU. | 17:53 |
edmondsw | oh, right, there's a spec for live resize but not currently implemented | 17:57 |
edmondsw | and yes, I think we would have to block live migrate for VMs with PCI devices attachd | 17:58 |
openstackgerrit | Chhavi Agarwal proposed openstack/nova-powervm master: iSCSI volume detach with no UDID https://review.openstack.org/576034 | 18:01 |
efried | edmondsw: Should we provide a CONF option denoting the fs path to the yaml config for passthrough? | 18:11 |
edmondsw | efried I'm not sure that's necessary... a well-known name in /etc/nova is probably fine | 18:20 |
efried | edmondsw: ack | 18:20 |
efried | edmondsw: Shall we call it device-passthrough.yaml ? | 18:20 |
efried | I wouldn't think a powervm- prefix is necessary, since it'll be on a PowerVM compute node. | 18:21 |
efried | And with any luck, we may become the gold standard that all drivers wind up using :) | 18:21 |
edmondsw | yeah, that'd be nice. I don't see a reason to prefix with powervm- | 18:24 |
edmondsw | maybe just passthrough.yaml? | 18:24 |
efried | shrug, okay. | 18:32 |
efried | edmondsw: Where (in what doc) do we describe our flavor extra_specs stuff? | 18:33 |
efried | I'm not seeing it in nova-powervm/doc/source anywhere. Do we maintain that somewhere separate? | 18:33 |
openstackgerrit | Eric Fried proposed openstack/nova-powervm master: Spec: Device Passthrough https://review.openstack.org/579359 | 18:38 |
efried | edmondsw: ^ | 18:38 |
edmondsw | efried I don't see any docs for our extra specs :( | 18:38 |
efried | boo | 18:39 |
edmondsw | agreed | 18:39 |
edmondsw | we need to fix that | 18:39 |
edmondsw | in the meantime, they're defined here in the code: https://github.com/openstack/nova-powervm/blob/7add56d035d98d58fd52d23b0a8beb04fcf78394/nova_powervm/virt/powervm/vm.py#L239 | 18:40 |
edmondsw | I'll add a TODO to the etherpad to include in docs | 18:40 |
edmondsw | efried hmm... we have a different list of supported extra_specs in-tree: https://github.com/openstack/nova-powervm/blob/7add56d035d98d58fd52d23b0a8beb04fcf78394/nova_powervm/virt/powervm/vm.py#L239 | 18:42 |
efried | yup, that's by design. I remember having to scrape through them when we started porting. | 18:42 |
edmondsw | these 3 | 18:43 |
edmondsw | https://github.com/openstack/nova-powervm/blob/7add56d035d98d58fd52d23b0a8beb04fcf78394/nova_powervm/virt/powervm/vm.py#L254-L256 | 18:43 |
edmondsw | do you remember why? | 18:43 |
efried | enable metric because we... don't. That would be a ceilometer thing? | 18:44 |
efried | I don't know what affinity check is. | 18:45 |
efried | And ppt ratio because... we don't care? | 18:45 |
edmondsw | I'm not satisfied with any of those answers... I'll try to dig into it | 18:45 |
edmondsw | I did find an existing todo for documenting extra specs | 18:45 |
edmondsw | mriedem had apparently commented on that before: https://review.openstack.org/#/c/391288/57/nova/virt/powervm/vm.py@339 | 18:46 |
efried | yup, just found that myself :) | 18:49 |
efried | I do remember we were always looking for things to subset when we moved in tree. | 18:49 |
efried | To reduce review footprint and avoid having to answer hard questions about "wtf is this and why do we need it?" | 18:49 |
edmondsw | sure, I just don't expect these would have been left out for those reasons | 18:50 |
edmondsw | ppt and affinity are very recent additions... I bet they were only added OOT just because that's what PowerVC uses and the folks that contributed them weren't thinking about IT | 18:51 |
edmondsw | may be the same for the other as well | 18:51 |
efried | ah, that could be. Looking... | 18:51 |
edmondsw | something we should be mindful of in future reviews | 18:51 |
efried | yup, those three came in after - at least after the in-tree port was proposed. | 18:53 |
efried | Yeah, well after. | 18:53 |
efried | https://review.openstack.org/#/c/498902/ | 18:54 |
efried | https://review.openstack.org/#/c/497977/ | 18:54 |
efried | https://review.openstack.org/#/c/555309/ | 18:54 |
openstackgerrit | Merged openstack/nova-powervm master: Fix coverage job https://review.openstack.org/580493 | 19:14 |
*** edmondsw has quit IRC | 23:26 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!