| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2022-07-20 | |||
| 13:43:24 | sean-k-mooney | yes | |
| 13:43:36 | sean-k-mooney | we cannot ignore a request for a feature | |
| 13:43:52 | sean-k-mooney | if the arch does not supprot it we must reject teh boot request | |
| 13:44:20 | sean-k-mooney | ricolin: this should not actully get that far really | |
| 13:44:31 | sean-k-mooney | the reason is that on those plathform we will not be reporting the traits | |
| 13:44:42 | sean-k-mooney | so we shoudl not get as far as generating the xml | |
| 13:45:21 | sean-k-mooney | but the current logic is incorrect we shoudl be raising an error if for some reason we are on a ppc64 host and we manage to get to this point in the virt driver | |
| 13:48:08 | sean-k-mooney | ricolin: the contract that nova provides is if you ask for a feature we will either fail to find a host that can provide that and set the instnace to error or we will boot a vm with that feature enabled | |
| 13:48:24 | sean-k-mooney | ricolin: so for this feature you need to provide that same guarentee | |
| 13:49:12 | ricolin | okay | |
| 13:49:21 | ricolin | I will add that error | |
| 13:49:41 | ricolin | you think on the libvirt driver works, or you're thinking elsewhere? | |
| 13:49:57 | ricolin | I mean place to add that check and raise error | |
| 13:50:35 | sean-k-mooney | so you shoudl raise the error in the libvirt driver | |
| 13:50:40 | sean-k-mooney | but in https://review.opendev.org/c/openstack/nova/+/844507/5 | |
| 13:50:54 | sean-k-mooney | you shoudl only report the iommu traits if you are on arm or x86 | |
| 13:51:21 | sean-k-mooney | that will prevent the libvirt hosts form being select if its say ppc64 or mips | |
| 13:51:51 | sean-k-mooney | so https://review.opendev.org/c/openstack/nova/+/844507/5/nova/virt/libvirt/driver.py#12102 need to check the architechtre. | |
| 13:54:31 | ricolin | got it | |
| 13:54:44 | ricolin | thanks sean-k-mooney | |
| 14:11:22 | gibi | sean-k-mooney: about the owner trait in the pci series. I can propose a standard trait for sure. I'm wondering if this will be really a compute capability traits as those are added to the root RP. I more thought about an RP type trait, expressing that this is a PCI RP. | |
| 14:56:42 | opendevreview | Balazs Gibizer proposed openstack/os-traits master: Add HW_PCI_DEVICE trait https://review.opendev.org/c/openstack/os-traits/+/850538 | |
| 14:56:55 | gibi | sean-k-mooney: I can go with the combination of ^^ and OWNER_NOVA | |
| 14:57:03 | gibi | that will express what I need | |
| 14:57:15 | gibi | later if we need a capability trait for scheduling then I will add one too | |
| 14:57:43 | sean-k-mooney | it does not really feel like a HW trait to me | |
| 14:58:03 | sean-k-mooney | more COMPUTE but i prefer HW over CUSTOM_ | |
| 14:58:25 | sean-k-mooney | gibi: its not an atirbute of the hardware its an atibute of the compute serivce | |
| 14:58:34 | sean-k-mooney | and its ablity to track pci device in this case via placment | |
| 14:58:56 | sean-k-mooney | if it was a properly of a specific pci device it woudl go under HW | |
| 14:59:10 | sean-k-mooney | is there a particalr reason you went with HW? | |
| 14:59:37 | gibi | I think we are still after two different things. You want to indicate that the compute can track PCI in Placement. I want to indicate that a specific RP represents a PCI device managed by Nova | |
| 14:59:56 | gibi | for the former I agree that is a compute capability trait | |
| 15:00:06 | sean-k-mooney | rigth but the later is not an atribute of the RP | |
| 15:00:19 | sean-k-mooney | its an atribute of the resouce clase tracked by one fo the inventires | |
| 15:00:21 | gibi | lets continue after the meeting... | |
| 15:00:32 | sean-k-mooney | sure | |
| 15:58:37 | gibi | I think I go your point. this is not a hw feature so the trait should not be under HW | |
| 15:58:52 | gibi | I will move it under COMPUTE | |
| 16:13:13 | opendevreview | Balazs Gibizer proposed openstack/os-traits master: Add COMPUTE_MANAGED_PCI_DEVICE trait https://review.opendev.org/c/openstack/os-traits/+/850538 | |
| 16:13:20 | gibi | sean-k-mooney: moved ^^ | |
| 16:52:39 | opendevreview | Balazs Gibizer proposed openstack/nova master: Basics for PCI Placement reporting https://review.opendev.org/c/openstack/nova/+/846187 | |
| 16:52:39 | opendevreview | Balazs Gibizer proposed openstack/nova master: Extend device_spec with resource_class and traits https://review.opendev.org/c/openstack/nova/+/846218 | |
| 16:52:40 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reject PCI dependent device config https://review.opendev.org/c/openstack/nova/+/846435 | |
| 16:52:40 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reject mixed VF rc and trait config https://review.opendev.org/c/openstack/nova/+/846436 | |
| 16:52:41 | opendevreview | Balazs Gibizer proposed openstack/nova master: Ignore PCI devs with physical_network tag https://review.opendev.org/c/openstack/nova/+/846219 | |
| 16:52:41 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reject devname based device_spec config https://review.opendev.org/c/openstack/nova/+/846466 | |
| 16:52:42 | opendevreview | Balazs Gibizer proposed openstack/nova master: Support [pci]device_spec reconfiguration https://review.opendev.org/c/openstack/nova/+/846470 | |
| 16:52:42 | opendevreview | Balazs Gibizer proposed openstack/nova master: Stop if tracking is disable after it was enabled before https://review.opendev.org/c/openstack/nova/+/847009 | |
| 16:52:43 | opendevreview | Balazs Gibizer proposed openstack/nova master: Allow enabling PCI tracking in Placement https://review.opendev.org/c/openstack/nova/+/850468 | |
| 16:52:44 | opendevreview | Balazs Gibizer proposed openstack/nova master: Move provider_tree RP creation to PciResourceProvider https://review.opendev.org/c/openstack/nova/+/850546 | |
| 18:10:16 | opendevreview | sean mooney proposed openstack/nova master: single pass instance info fetch in host manager https://review.opendev.org/c/openstack/nova/+/623558 | |
| 19:29:31 | opendevreview | Amit Uniyal proposed openstack/nova master: add regression test case for bug 1978983 https://review.opendev.org/c/openstack/nova/+/849104 | |
| 21:09:08 | opendevreview | Rico Lin proposed openstack/nova master: Add locked_memory extra spec and image property https://review.opendev.org/c/openstack/nova/+/778347 | |
| 21:09:09 | opendevreview | Rico Lin proposed openstack/nova master: libvirt: Add vIOMMU device to guest https://review.opendev.org/c/openstack/nova/+/830646 | |
| 21:09:10 | opendevreview | Rico Lin proposed openstack/nova master: Add traits for viommu model https://review.opendev.org/c/openstack/nova/+/844507 | |
| #openstack-nova - 2022-07-21 | |||
| 07:27:27 | gibi | o/ | |
| 07:27:33 | bauzas | \o | |
| 07:29:06 | Uggla | o/ | |
| 07:48:49 | bauzas | hmmmm, Kendall is asking me a few questions about the PTG | |
| 07:49:01 | bauzas | I guess this email was sent to all PTLs | |
| 07:49:27 | bauzas | tl:dr: the Foundation wants to invite ops at the PTG, which I think is an excellent proposal | |
| 07:49:55 | bauzas | but now, in order to make this happen, we somehow have to showcase a bit and think how we could make their trip valuable | |
| 07:50:43 | bauzas | if people are willing to, we can brainstorm a bit about those 4 questions she sent | |
| 07:50:55 | bauzas | 1/ what do we want to hear from ops at the PTG | |
| 07:51:08 | bauzas | 2/ what topics are on our PTG agenda (yet) | |
| 07:51:24 | bauzas | 3/ what are the goals for the Columbus PTG ? | |
| 07:52:08 | bauzas | 4/ what contributors think about an extra day on Friday if there is an Ops sessions day ? | |
| 07:53:57 | MichielPiscaer[m] | I as ops person, I'm planning to got to the PTG. I think the main question is what do you want to know from me as operator. | |
| 08:02:07 | gibi | bauzas: good questions. I think we should fire up an etherpad to collect the answers from the team | |
| 08:02:40 | bauzas | gibi: this was my approach | |
| 08:03:17 | bauzas | II don't want to hold that much to reply to diablo_rojo | |
| 08:03:41 | bauzas | so ideally, I'd appreciate if I could reply by tomorrow (hence that not being discussed at the nova meeting) | |
| 08:04:11 | bauzas | also, Kendall is asking whether we agree on publishing our responses into some superuser blog | |
| 08:04:38 | bauzas | MichielPiscaer[m]: oh, thanks for your feedback, extremely valuable | |
| 08:05:09 | bauzas | MichielPiscaer[m]: in general, we appreciate some ops presence in the room for challenging the usecases | |
| 08:06:12 | bauzas | during the last 3 virtual PTGs, we didn't really went into deepdive design sessions, so I feel ops couldn't be that disconnected from the talks | |
| 08:08:34 | gibi | bauzas: #1) new use cases, pain points, general info about what they are using (services, versions, configuration) | |
| 08:09:03 | gibi | bauzas: #2) I will definielty talk about PCI in placement :) but not much else on my agenda so far | |
| 08:09:35 | gibi | bauzas: #3) plan the AA (and BB cycle) | |
| 08:09:56 | bauzas | I'm just creating the etherpad with my notes | |
| 08:09:57 | gibi | bauzas: #4) I'm happy to have an extra day with Ops | |
| 08:10:01 | bauzas | me too | |
| 08:10:17 | gibi | ack, I will update the etherpad too | |
| 08:12:14 | sean-k-mooney | gibi: i am hoping it will be functinal if but not nessisarly on by default in AA | |
| 08:12:47 | bauzas | https://etherpad.opendev.org/p/nova-ptg-columbus-ops-presence | |
| 08:13:04 | bauzas | gibi: add your points to the agendaz | |
| 08:13:05 | sean-k-mooney | i would potentialy like to see the mdev support evloved to build in pci in placment and the pci track ro move to the resouces table like pmem overtime | |
| 08:13:05 | gibi | sean-k-mooney: yepp I think we still have a chance to have the non neutron part in Zed, and then the neutron part in AA | |
| 08:13:30 | sean-k-mooney | gibi: ack i would personally hold enabling by default until we have parity | |
| 08:13:36 | sean-k-mooney | so likely BB | |
| 08:13:58 | bauzas | tbc, this is a bit premature to know what the topics will be | |
| 08:14:02 | gibi | sean-k-mooney: I agree with keeping off by default until we have full parity | |
| 08:14:56 | sean-k-mooney | bauzas: all topic sure but we will have some in the back of our minds | |
| 08:15:15 | bauzas | sean-k-mooney: then add them to the etherpad I mentioned | |
| 08:15:27 | bauzas | and I'll summarize those | |
| 08:15:43 | sean-k-mooney | bauzas: we ususally have a corss project day on monday should we make that ops and cross project | |
| 08:15:54 | bauzas | keeping in mind most ops are running pre-Train :) | |
| 08:16:09 | bauzas | so most of our semantics are mostly unknown by them :) | |
| 08:16:32 | bauzas | sean-k-mooney: looks like there will be a ops day on Friday | |