Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-04-05
15:43:10 mriedem stephenfin: wouldn't that information be in the port's host binding details?
15:46:25 stephenfin mriedem: Possibly, but we don't do any of that stuff until we land on a host, far as I can tell
15:47:01 stephenfin Only thing we do do (heh) is prepare a PCIRequest object for any SR-IOV requests
15:48:04 stephenfin bauzas: Sorry to hear it :(
15:48:34 bauzas stephenfin: no worries, it was expected
15:48:45 sean-k-mooney mriedem: that info is not in the port until after binding
15:49:07 sean-k-mooney stephenfin: i dont think it was in a seperat spec but it was part of the bandwith based scheduling spec
15:49:41 stephenfin sean-k-mooney: Hmm, OK. Was hoping I could steal some ideas from said spec
15:50:05 sean-k-mooney the intent was to use traits for neturon extention or backend to select the backend. the trati would be added to the neutron port resouce reuest
15:50:34 stephenfin So you'd be querying placement?
15:50:40 sean-k-mooney stephenfin: well you can proably steal them from the bandwith spec.
15:50:46 sean-k-mooney stephenfin: yes
15:51:05 sean-k-mooney or rather its addtional info that is included with nova queries placement
15:51:12 stephenfin Darn, it all comes back to placement
15:52:14 sean-k-mooney stephenfin: i did have a proposal to do this with a neutron api for network backend capablities in mitaka before placement but that never went anywhere outside intel
15:52:37 stephenfin Yeah, so NUMA stuff. I naturally want to be able to do this stuff in the scheduler but I don't have enough information at that point to make a decision
15:52:52 openstackgerrit Matt Riedemann proposed openstack/nova master: doc: add a link in the install guides about configuring neutron https://review.openstack.org/559115
15:53:03 stephenfin I can figure out what network I have but then I need to map that network to a given NUMA node
15:53:39 stephenfin and that information lives on each individual host, either as a configuration option or something else
15:53:55 sean-k-mooney stephenfin: well you can have enough info in the schduler if you populated it in the host state object. placement does not know about it however unlees the compute agent puts it in placement too
15:54:13 stephenfin so it looks like I'm going to have to pass it back to the scheduler via the host state object, which I'm not sure I want to do
15:54:21 stephenfin sean-k-mooney: yup, exactly
15:54:28 sean-k-mooney stephenfin: i had assumed you were going to addd the network stuff to the numa topology blob
15:55:24 openstackgerrit Matt Riedemann proposed openstack/nova master: doc: add a link in the install guides about configuring neutron https://review.openstack.org/559115
15:55:49 sean-k-mooney so the compute agent would read the config opptions and populated the numa info for the backend into the numatopology blob then the numa topology filter which gets that blob in the host_State object can make desions about the host
15:57:02 openstackgerrit Matt Riedemann proposed openstack/nova master: Update the cells FAQs and scheduler maintenance docs. https://review.openstack.org/556459
15:57:43 stephenfin sean-k-mooney: Hmm, I hadn't considered that. I'd been adding additional fields and then got stuck because I had a network from the user and physnet name from the host
15:57:54 sean-k-mooney stephenfin: your only other option bar modleing it in placement which is the long term answer is to not scheduler based on the numa constriats and retry if you cannont fit the insance on the compute node that was selected
15:57:55 stephenfin *I had considered that but...
15:58:41 sean-k-mooney stephenfin: well you have more then the network
15:58:58 sean-k-mooney your stuff assumed that the vm was booted with a nuton port
15:59:10 stephenfin so network and port IDs
15:59:11 stephenfin ?
15:59:27 sean-k-mooney in that case we already pull back the full neutron port object to check if its an sriov port
15:59:46 sean-k-mooney so you can ask neutron for the physnet and put that in the instance request object
16:00:04 sean-k-mooney you can also determin if its a tunneled/routed port
16:00:31 stephenfin sean-k-mooney: Yup, I've been charting that this afternoon :D http://paste.openstack.org/show/718499/
16:01:06 sean-k-mooney stephenfin: thats how the nic feature based schduling works :)
16:01:27 stephenfin ralonsoh_'s patches?
16:01:31 sean-k-mooney yep
16:01:46 stephenfin (I've been meaning to re-review those but they were failing CI last I checked)
16:02:05 sean-k-mooney we extend the pci requst spec object to carry the feature flags
16:02:59 sean-k-mooney stephenfin: i belive the bandwidth spec was suggestion adding an arry of networkRequstsSepcs to the instance request object
16:03:37 stephenfin I need to reread that one too so. That's exactly what I'd done
16:04:01 sean-k-mooney stephenfin: oh and i started rebasing rodoflos patches yesterday
16:04:18 stephenfin Only, like I said, it seemed useless to me because I didn't want to be querying neutron for more information from the scheduler
16:04:19 sean-k-mooney i hope to have them back up tomrowo or monday
16:04:31 stephenfin sean-k-mooney: Sounds good. I can review once they're done
16:04:35 stephenfin *ready
16:05:22 stephenfin sean-k-mooney: I'll push up what I have probably some time tomorrow
16:05:33 stephenfin Try not to laugh too hard. It's strewn with TODOs :D
16:05:34 sean-k-mooney well i broke them more in the refactor. once they are back passing unuit test i will need to do some more testing with real hardware so they wont be "ready" untill later next week but they can still be reviewed
16:05:54 stephenfin Yeah, that's fine. I can do some basic validation too
16:06:02 stephenfin ...given that I actually have hardware again
16:06:31 sean-k-mooney well im currently doing the rebase work in a vm so need to deploy on real hard ware to test after that is done.
16:12:16 mriedem efried: imacdonn: i've updated https://review.openstack.org/#/c/554577/ to depend on your nova change
16:13:34 efried mriedem: cool
16:30:05 mriedem kashyap: holy wow everything volume-related failed in your patch http://logs.openstack.org/84/534384/24/check/tempest-full/c7c0cbe/
16:30:09 mriedem congratulations on breaking cinder
16:31:03 kashyap mriedem: I suppose you're referring to: https://review.openstack.org/#/c/558783/
16:31:05 mriedem it's not related to your patch, just looks like c-vol is having a very bad time http://logs.openstack.org/84/534384/24/check/tempest-full/c7c0cbe/controller/logs/screen-c-vol.txt.gz?level=TRACE
16:31:06 kashyap ?
16:31:16 mriedem no
16:31:21 mriedem https://review.openstack.org/#/c/534384/
16:31:43 kashyap Ah, that one
16:32:25 kashyap mriedem: I go do a plain 'recheck', or does it require a more deeper incantation?
16:32:32 mriedem just recheck it
16:32:43 kashyap Done
16:33:13 kashyap mriedem: Too eager to jump the gun, eh :P
16:34:38 openstack Launchpad bug 1758278 in OpenStack Compute (nova) "disk_available_least become a negative value unexpectedly" [Undecided,New]
16:34:38 melwitt triaging this bug about disk_available_least going negative for an image-based instance with 0 GB disk flavor. proposal is to reject such requests with 400 (which would require a microversion, I think). additional thoughts welcome https://bugs.launchpad.net/nova/+bug/1758278
16:36:14 kashyap dansmith: Thanks for the fast-approve, and the eagle eyes, as usual.
16:38:41 jaypipes efried, pls see my response on https://review.openstack.org/#/c/556873
16:38:52 efried jaypipes: ack
16:39:27 mriedem dansmith: if you're done with your meeting, this will close out the cell-disable bp https://review.openstack.org/#/c/556459/
16:39:32 mriedem just the final docs patch
16:51:20 openstack Launchpad bug 1761197 in OpenStack Compute (nova) "Not defined keypairs in instance_extra cellsV1 DBs" [Undecided,New] - Assigned to Surya Seetharaman (tssurya)
16:51:20 melwitt can someone remind me, in this last comment from belmiro, did he mean cells v2 instead of cells v1 or? https://bugs.launchpad.net/nova/+bug/1761197
17:01:42 openstackgerrit Arvind Nadendla proposed openstack/nova master: Update ImageMetaProp object to expose traits https://review.openstack.org/557795
17:02:29 arvindn051 dansmith: jaypipes: updated code per comment(changed name of the field)
17:14:35 jaypipes dansmith: you good with it, I'm good with it.
17:14:56 dansmith I'm under a pile of other stuff right now,
17:14:58 dansmith hence me ignoring pings
17:14:59 dansmith sorry
17:15:02 dansmith will look later
17:15:59 jaypipes dansmith: no worries. I left a +2 on it.
17:26:20 sean-k-mooney jaypipes: just thinking back over your cpu spec and some of the other granualr requests discussions. have we finalised on the fact that numbered request groups will be guraenteed to be form different RPs
17:29:53 efried sean-k-mooney: Nobody has proposed a spec change
17:31:18 sean-k-mooney efried: hum ok in the current form if they are in the same request group are all resouces in that group required form the same subtree?
17:31:56 sean-k-mooney efried: ill read the sepc to confim just tought i would ask if you knew off the top of your head
17:32:09 efried if in the same numbered request group, they'll land in the same provider
17:32:38 sean-k-mooney efried: same provider or "same proviers or one of its childern"
17:32:38 efried in the unnumbered group, can be spread around the tree
17:33:56 openstack Launchpad bug 1761197 in OpenStack Compute (nova) "Not defined keypairs in instance_extra cellsV1 DBs" [Undecided,New] - Assigned to Surya Seetharaman (tssurya)
17:33:56 melwitt tssurya: in this bug in belmiro's last comment, did he mean keypair isn't in instance_extra with cells v2? the comment says "cells v1" https://bugs.launchpad.net/nova/+bug/1761197
17:34:17 efried sean-k-mooney: same provider. The only thing we have today (impl or design) that uses children is the concept of grouping as a tree. Nothing in placement deals with hierarthies or branches or subtrees etc
17:34:57 efried hierarchies, that is
17:35:02 melwitt fwiw, I checked a recent devstack of mine and see keypairs in instance_extra in the nova_cell1 database, so if there's a bug in ocata, it looks like it got fixed somehow later
17:35:27 melwitt (assuming cells v2 was what he meant)
17:36:19 sean-k-mooney efried: so you can never have a numbered resouce group containig CPUs and virtual fucntions and find a host as they will be from diffrent RPs in a subtree of a compute host

Earlier   Later