Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-04-12
15:44:56 gameon sahid: Thanks again for your help - I have resolved the issue. It turns out Broadwell doesn't have all Haswell features and vice versa. So I've used IvyBridge
15:45:27 efried cfriesen: Although it amounts to the same thing, really.
15:45:35 cfriesen gameon: what was in haswell that's not in broadwell?
15:45:38 efried cfriesen: So yeah, edleafe there's your counterargument.
15:46:16 sahid gameon: cool :)
15:47:31 edleafe efried: I would much prefer modifying the provider info than the a-c
15:47:58 efried edleafe: But our problem isn't picking a provider. It's picking an allocation request.
15:48:07 gameon cfriesen: this was generated by running virsh capabilities - I the features reported by 'cat /proc/cpuinfo' are the same... http://paste.openstack.org/show/719064/
15:48:18 efried edleafe: ...from among possibly many for a given provider.
15:48:41 efried edleafe: Besides, we're not talking about modifying anything about the format of the response.
15:48:48 gameon cfriesen: happy to hear alternative solutions to setting model to IvyBridge there's a better way of allowing LM between these set of hosts
15:49:23 efried edleafe: But we are talking about introspecting that payload. Which nothing in the doc implies we should avoid...
15:49:50 edleafe efried: but it also doesn't guarantee that the payload structure will never change
15:50:07 edleafe "Introspect at your own risk" :)
15:50:31 cfriesen gameon: according to my libvirt cpu model definitions Broadwell should be a superset of Haswell.
15:50:46 bauzas edleafe: just a bit of explanation
15:50:48 cfriesen gameon: and you might want to use the "noTSX" versions
15:51:09 bauzas edleafe: we discussed since 3 hours with efried about the possible implementation for NUMA
15:51:19 openstackgerrit Matt Riedemann proposed openstack/nova stable/pike: Don't persist RequestSpec.retry https://review.openstack.org/560146
15:51:32 bauzas edleafe: the problem is that the NUMA filter makes some assumptions that are not supported by the numbered request groups feature
15:51:48 gameon cfriesen: I'll give it a go, let's see. So set it to Broadwell or Haswell do you think?
15:51:58 bauzas edleafe: the one in particular is that it checks whether the host has the exact same topology than the asked guest one
15:52:07 bauzas edleafe: which seems weird to me, but meh
15:52:44 cfriesen gameon: interesting that it doesn't report your Haswell as a Haswell.
15:52:52 bauzas edleafe: accordingly, that means that placement will return some allocation requests that are valid from his point of view, but not from the filter's PoV
15:53:17 bauzas edleafe: hence the need of some post-filtering that would only assure the same behaviour
15:53:20 bauzas but
15:53:21 gameon cfriesen: Yes I noticed this, reported as Nehalem which is a bit puzzling...
15:53:23 bauzas there is a but
15:53:25 cfriesen gameon: According to what you've got there I'd use Nehalem, unless you need the newer features.
15:53:57 bauzas edleafe: one other option could be to consider that placement should return the exact same candidates as the ones that the filter would agree
15:54:10 bauzas edleafe: in that case, it would mean some new query param for Placement
15:54:21 bauzas thoughts ?
15:55:05 gameon cfriesen: Nehalem works :)
15:56:27 efried edleafe: The way the doc is written, it *is* guaranteeing that the format won't change (without a new microversion).
15:56:53 efried which is as it should be.
15:57:10 edleafe efried: about to run off to a meeting
15:57:17 efried I've never understood the reasoning behind that payload needing to be opaque, btw.
15:57:50 edleafe efried: all it was supposed to be was something you could send back to allocate/claim
15:57:53 cfriesen bauzas: it seems to me that we have two options. give the filters access to the allocation candidates so they can rule out ones they don't like, or have enough flexibility in placement that we can ensure we never get back invalid candidates.
15:58:03 edleafe it could have been a hash, or a uuid, or...
15:58:22 efried edleafe: And you're supposed to pick one based on... what?
15:58:32 cfriesen edleafe: for that to work we need enough flexibility in what we can request from placement to ensure that all the candidates are valid
15:58:33 edleafe now we have nested, which brings in multiple a-cs per rp
15:58:35 melwitt efried: when you get a chance, wanna write some notes on how the update-provider-tree runway review went at L118? https://etherpad.openstack.org/p/nova-runways-rocky
15:58:47 efried melwitt: ack
15:58:51 edleafe efried: you picked a host, and found the matching a-c
15:59:08 efried edleafe: There were several a-cs for that host.
15:59:13 efried edleafe: How did I pick one?
15:59:19 edleafe now there isn't a 1:1 host:a-c relationship with nesting
15:59:22 cfriesen efried: if they were all actually valid it wouldn't matter
15:59:37 efried cfriesen: Yup, that's the big IF.
16:00:05 efried cfriesen: For this use case, we either implement new logic in placement, or they're *not* all valid and we can't just pick one at random.
16:00:14 cfriesen efried: yes, agreed
16:00:43 efried edleafe: There was never a 1:1 host:a-c relationship.
16:05:01 openstackgerrit Matt Riedemann proposed openstack/nova stable/ocata: Don't persist RequestSpec.retry https://review.openstack.org/560955
16:06:24 bauzas efried: just for the context, they're not all valid because of some specific implementation details of the filter that I don't particularly like
16:06:53 efried melwitt: done
16:06:58 bauzas adding more debt to either placement or the filters looks terrible to me
16:07:06 openstackgerrit Matt Riedemann proposed openstack/nova stable/ocata: Don't persist RequestSpec.retry https://review.openstack.org/560167
16:07:18 melwitt efried: thanks
16:07:45 bauzas efried: from a placement perspective, we can ask for a query that would *shard* resources between children, but that's the only trade-off I'd make
16:08:18 bauzas efried: if we implement such thing, then we wouldn't need to pass the candidates down to the filters
16:08:22 efried bauzas: It's starting to sound like that might be the "easier" option
16:08:37 bauzas not the easier
16:08:43 efried bauzas: And as I've said, we know we're going to want that logic in placement eventually regardless.
16:08:46 bauzas the less debtful
16:08:49 efried bauzas: So this might as well be the motivation.
16:08:51 efried Yeah
16:09:19 bauzas ok, I'll amend my spec accordingly
16:09:27 edleafe efried: sorry, in the API-SIG meeting
16:09:30 edleafe efried: https://github.com/openstack/nova/blob/master/nova/scheduler/manager.py#L154-L160
16:09:49 edleafe That creates one a-c per rp
16:10:00 melwitt cdent: the placement-forbidden-traits blueprint has been added to a review runway. please ack if the next two weeks work for you for quick iteration on review
16:10:25 edleafe we always just grab the first one of that list
16:11:13 cdent melwitt: thanks, it does
16:11:24 melwitt k, great
16:11:28 efried edleafe: It creates a *list* of allocation requests per rp_uuid
16:11:44 efried edleafe: ...by introspecting the payload, by the way :P
16:13:41 openstack Launchpad bug 1739593 in OpenStack Security Advisory "Swapping encrypted volumes can lead to data loss and a possible compute host DOS attack (CVE-2017-18191)" [Undecided,Incomplete]
16:13:41 mriedem lyarwood: for https://review.openstack.org/#/c/543569/ - do we want a security reno for https://bugs.launchpad.net/nova/+bug/1739593 and CVE-2017-18191?
16:13:45 mriedem i see the ossa isn't published
16:14:07 melwitt jichen: the z/VM driver series has been added to a review runway. I know you have been active on the patches already but please let us know if there are any problems with the next two weeks for quick iteration on review
16:14:16 edleafe efried: this is the part that I was referring to: https://github.com/openstack/nova/blob/master/nova/scheduler/filter_scheduler.py#L212-L217
16:14:25 edleafe "information in the provider summaries"
16:17:36 efried edleafe: Yeah, I get that we can do *some* weighing/filtering based on the provider summaries; but that still only gets us down to the list of allocation requests for a given host. It doesn't help us pick among those.
16:20:45 mriedem lyarwood: sounds like the ossa is blocked until the stable/ocata patch is up
16:20:54 mriedem i'm +1 on the stable/pike change now if you want to start on the stable/ocata backport
16:27:02 openstackgerrit Brianna Poulos proposed openstack/nova master: Implement certificate_utils https://review.openstack.org/479949
16:28:25 cfriesen mriedem: lyarwood: I've got https://review.openstack.org/#/c/560690/ up for the stable/pike backport, but there's a complication in that Pike treats the encryption stuff a bit differently. Wondering how you want to handle it. I wrote it up in the review.
16:29:04 mriedem cfriesen: heh, see https://review.openstack.org/#/c/543569/
16:32:13 mriedem cfriesen: comments inline
16:32:46 openstackgerrit Brianna Poulos proposed openstack/nova master: Add trusted_image_certificates to REST API https://review.openstack.org/486204
16:34:42 mriedem cfriesen: oh yeah reading https://review.openstack.org/#/c/460243/ i see the problem kind of,
16:34:51 mriedem that was really a frankenstein of a patch, and should have been split up
16:36:02 mriedem https://review.openstack.org/#/c/460243/16/nova/virt/libvirt/driver.py@1453 specifically
16:36:15 mriedem if that's its own bug, we'd have to backport separately before your change, but would need to talk to lyarwood
16:36:47 mriedem i don't know if that applies before the changes to _disconnect_volume though
16:36:53 mriedem if not, don't worry about it
16:38:43 cfriesen mriedem: If we wanted to backport "proper" handling of encrypted volumes in the error case it could be done in a separate patch, I don't think the ordering really matters. For now I'll rework the backport to go off the stable/queens one.

Earlier   Later