Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-05-31
15:52:04 dansmith are you agreeing with that or asserting something else?
15:52:14 mriedem dansmith: you and i are in agreement,
15:52:36 mriedem stephenfin found that we pass requested_networks down to compute also, which will have the same type of informatoin in them so we could use that, and i'm asserting we shouldn't b/c it's confusing
15:52:48 dansmith agree
15:53:06 mriedem stephenfin: numa_fit_instance_to_host doesn't modify the limits dict today does it?
15:54:32 stephenfin mriedem: Narp. _numa_fit_instance_cell consumes any NUMATopologyLimits objects contained therein but doesn't modify it
15:54:38 mriedem stephenfin: i don't think that numa_fit_instance_to_host should modify the limits dict, the scheduler filter should do that
15:54:58 mriedem i would basically like to keep all of this as decoupled as reasonably possible
15:55:04 mriedem because right now it's a tightly coupled mess
15:55:33 mriedem and it's like the last part of nova i'm not really familiar with, and don't really ever want to be if i can help it :)
15:56:07 stephenfin By scheduler filter you mean the NUMATopologyFilter?
15:56:10 mriedem yes
15:56:17 mriedem the thing that already puts the numa limits in the limits dict today
15:56:26 mriedem the scheduler filters are the only things that modify the limits dict
15:56:32 mriedem we should maintain that pattern
15:56:56 openstackgerrit Merged openstack/nova stable/ocata: Add ssbd and virt-ssbd flags to cpu_model_extra_flags whitelist https://review.openstack.org/570514
15:57:02 stephenfin We're still going to need some way to get that info into the filter though
15:57:17 stephenfin So we'd still be modifying RequestSpec
15:57:20 mriedem the request spec will have the requested_networks right?
15:57:26 mriedem yes i know and agreed with that already
15:58:22 mriedem api -> modify reqspec.requested_networks like pci requests -> scheduler -> numa filter + modify limits -> conductor -> compute -> RT (claim from limits) -> allocate_for_instance -> update instance info cache with network physnet and tunneled info -> driver.spawn
15:58:24 mriedem done!
15:59:07 stephenfin Oh, so I still have the RequestSpec changes which means for the scheduler I'll just be undoing this change https://review.openstack.org/#/c/564448/3/nova/virt/hardware.py@1609
15:59:13 stephenfin (where I add 'network_requests'
15:59:28 stephenfin ...and instead consuming 'network_requests' from limits
15:59:53 belmoreira mriedem: I don't have the logs for https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L531
16:00:01 stephenfin Then I can use that for the call to 'instance_claim'
16:00:02 stephenfin Gotcha
16:00:06 mriedem stephenfin: i don't really know what that code is doing, so hard for me to say
16:00:17 stephenfin mriedem: I'm just referring to the function signature
16:00:32 mriedem stephenfin: i'm missing the context at which numa_fit_instance_to_host gets called
16:00:34 mriedem i know the numa filter calls it,
16:00:38 mriedem but the RT does also correct?
16:01:01 mriedem well, _test_numa_topology in the claims code, called by the RT
16:01:04 stephenfin mriedem: Yeah, that's here https://review.openstack.org/#/c/564449/3/nova/compute/claims.py
16:01:26 stephenfin I could undo most of those changes
16:01:30 mriedem right so i think you don't need to pass requested_networks in there, just get it from the limits
16:01:36 stephenfin Exactly
16:01:44 stephenfin OK, that clears things up massively. Cheers :)
16:01:51 mriedem do you need to pass anything into numa_fit_instance_to_host from the numa filter?
16:02:06 mriedem stephenfin: heh, yeah, you were over-complicating this it sounds like
16:02:26 stephenfin I'll need to set the 'network_requests' field in limits but that's it
16:02:36 mriedem remember, at any given point in any given location in nova, we have at least 3 random dicts of stuff you can leverage for master hackery
16:02:55 stephenfin mriedem: Yeah, PCI requests probably wasn't the best pattern to copy
16:03:06 mriedem definitely not
16:03:12 mriedem i cringe whenver i see that
16:08:45 openstackgerrit Merged openstack/nova master: Update overriden to overridden https://review.openstack.org/571418
16:10:05 stephenfin jaypipes: What's your thoughts on all the above?
16:11:02 mriedem tssurya: i've added some detailed comments to https://bugs.launchpad.net/nova/+bug/1771806/comments/3
16:11:03 stephenfin jaypipes: tl;dr: We're going to stuff the NetworkRequestList object into the limits dictionary like we do for NUMATopologyLimits https://github.com/openstack/nova/blob/master/nova/scheduler/filters/numa_topology_filter.py#L102
16:11:03 openstack Launchpad bug 1771806 in OpenStack Compute (nova) "Ironic nova-compute failover creates new resource provider removing the resource_provider_aggregates link " [Medium,Confirmed] - Assigned to Surya Seetharaman (tssurya)
16:11:27 mriedem stephenfin: i didn't think it would be that entire NetworkRequestList object
16:11:42 mriedem i thought it would be actual specific limits things
16:12:04 stephenfin mriedem: A new object so?
16:12:19 mriedem maybe, i'm not sure what your data structure needs to look like for the limits stuff
16:12:37 mriedem but i'm pretty sure it could certainly be stripped down from the full NetworkRequestList right?
16:12:50 stephenfin It can, yeah. I could reuse NUMANetworkInfo https://review.openstack.org/#/c/564439/3/nova/objects/numa.py
16:13:58 mriedem maybe, whatever it is will go into the SchedulerLimits object https://github.com/openstack/nova/blob/master/nova/objects/request_spec.py#L774
16:14:20 tssurya mriedem: looking
16:14:35 mriedem tssurya: just the same stuff i said in irc, but with detailed links
16:15:47 oomichi alex_xu: are you still online now? happy if you take a look at https://review.openstack.org/#/c/571335
16:19:19 melwitt mriedem: I was thinking queens release will be a minor version bump because of this, do you agree? https://git.openstack.org/cgit/openstack/nova/commit/?h=stable/queens&id=45331d3ea185d08687e9b24e716f3d5411210874
16:21:07 mriedem melwitt: umm,
16:21:29 mriedem i personally don't really see the need for a minor version bump, i don't know exactly what that will signal since we already have a release note
16:21:37 melwitt api change? I wasn't sure
16:21:51 mriedem about the only time i've seen stable branch minor version bumps is when we have to adjust dependencies,
16:21:52 melwitt okay
16:21:56 mriedem or maybe with a new db migration
16:22:04 mriedem the api change is to fix broken behavior
16:22:13 mriedem so 409 rather than screw up your data
16:23:00 melwitt okay, I've got the bugfix version bump proposed at https://review.openstack.org/571494
16:24:25 mriedem i'd like to get https://review.openstack.org/#/c/571425/ in if we can
16:24:28 mriedem the patch below it needs a +W
16:24:37 mriedem that was related to the nova-lvm regression we backported
16:24:47 mriedem bauzas: can you hit this? https://review.openstack.org/#/c/571424/1
16:24:56 melwitt k
16:30:50 melwitt mriedem: will do similar for pike and ocata. we also have this for pike and ocata that just need +W https://review.openstack.org/561850
16:39:38 melwitt release proposed for pike https://review.openstack.org/571521 and ocata https://review.openstack.org/571522
16:40:49 mriedem jaypipes: bauzas: thanks for the comments on the heal_allocations change, i've replied https://review.openstack.org/#/c/565886/
16:41:33 mriedem melwitt: ok i'll look at those backpots
16:41:34 mriedem *ports
16:42:20 melwitt thanks
16:42:25 openstackgerrit Stephen Finucane proposed openstack/nova-specs master: Add 'numa-aware-vswitches' spec https://review.openstack.org/541290
16:51:10 mriedem melwitt: +W on both of those backports
16:51:29 melwitt sweet thanks
16:52:23 melwitt need to round up some reviews for the skip fetching virtual size patches
16:52:55 melwitt something in this one isn't passing unit test jobs tho https://review.openstack.org/571432
16:55:14 mriedem yeah i already talked with lee about it
16:55:35 mriedem i can call in stable maint core ringers if needed on the rest
16:55:46 mriedem i.e. mtreinish
17:02:40 mriedem easy libvirt driver change for the trusted vf series https://review.openstack.org/#/c/458514/
17:02:40 jaypipes mriedem: "The missing consumer records" <-- worst Nancy Drew title ever.
17:02:58 mriedem jaypipes: it was a dark and stormy night and the consumer records were....MISSING
17:03:01 mriedem duh duh duh
17:03:17 jaypipes :)
17:35:37 openstackgerrit Matt Riedemann proposed openstack/nova master: Match ComputeNode.uuid to ironic node uuid in RT https://review.openstack.org/571535
17:41:21 mriedem is artom around today?
17:46:10 openstackgerrit karim proposed openstack/nova master: Handle rebuild of instances with image traits https://review.openstack.org/571539
17:46:10 melwitt he was around earlier this morning but logged off about an hour ago
17:55:45 openstackgerrit Matt Riedemann proposed openstack/nova master: metadata: add vf_trusted field to device metadata https://review.openstack.org/566344

Earlier   Later