Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2026-04-17
14:55:57 winiciusallan[m] folks, I saw a lot of discussion on the mail list about a spot instance service in Nova, some people have even start implementing, but this wasn't go further
14:56:04 winiciusallan[m] CERN has a external service for this https://gitlab.cern.ch/cloud-infrastructure/aardvark
14:56:29 winiciusallan[m] do you know by chance if having this kind of service integrated into Nova is interest for the community?
15:00:08 winiciusallan[m] i'm thinking of implementing this as part of my final thesis. this will make other research regarding billing and charge strategies easier to evaluate
15:14:06 opendevreview Merged openstack/nova master: tox: Remove override of install_command https://review.opendev.org/c/openstack/nova/+/984918
15:14:48 opendevreview Merged openstack/nova master: docs: Update development-environment guide https://review.opendev.org/c/openstack/nova/+/984956
15:42:51 sean-k-mooney winiciusallan[m]: we decieed that it should not be in nova
15:43:24 sean-k-mooney we reivwed the sepec and went back and forth but decieed it shoudl live as its own thing external to nova
15:43:44 sean-k-mooney winiciusallan[m]: i have tought that this could live in watcher in the future
15:43:52 sean-k-mooney there is also a host based version in blazar today
15:44:23 sean-k-mooney but i think watcher can facilatate a more granular approch similar to the orgianl concept
15:49:38 winiciusallan[m] sean-k-mooney: yeah, thinking about watcher it looks more natural considering the architecture of the service
15:50:32 winiciusallan[m] I dropped a message on scientific-sig asking if people have use cases or gaps that they would like to improve
15:51:14 winiciusallan[m] one from NASA talked about GPUs being expensive and it might be a use case for them
15:51:22 sean-k-mooney i dont know if i ever wrote up a spec for it there. but the idea was to annotate the isntance via the flavor or instance metadta wiht lifecycle:premtable=ture
15:51:45 winiciusallan[m] others from CERN and nectar-rc use aardvak
15:52:05 sean-k-mooney then have a new goal for permation and 1-n stragies for how to dop it
15:52:38 winiciusallan[m] sean-k-mooney: +1 with this approach. this is how the first spec was written on nova
15:53:26 winiciusallan[m] probably different strategies for how to find the best instance to preempt and also being possible to extend a custom strategy
15:53:26 sean-k-mooney so the firction with doing this is nova is https://docs.openstack.org/nova/latest/contributor/project-scope.html#no-more-orchestration and https://docs.openstack.org/nova/latest/contributor/project-scope.html#no-more-orchestration
15:53:31 winiciusallan[m] sounds good
15:54:01 sean-k-mooney we have basicly said we do not want to build complex orchstration logic in nova or optimise for batch processing
15:54:16 sean-k-mooney but we want to make sure that higher level service can express the policy on top
15:54:35 sean-k-mooney which is where aardvark or watcher can play a role
15:55:33 sean-k-mooney winiciusallan[m]: ya so there are a lot of intersting way to decie which instnace to preempt and what that means (either delete or shelve)
15:55:42 sean-k-mooney for example delete oldest first
15:55:50 sean-k-mooney delete bigest first
15:56:02 winiciusallan[m] best fit and so on...
15:56:17 sean-k-mooney delete instnace on expensive hosts or where there is pressure based on mettrics
15:56:31 sean-k-mooney so watcher with its plugable goals and stragies
15:56:38 sean-k-mooney woudl allow you to experiment with those
15:58:34 sean-k-mooney winiciusallan[m]: https://redhat.atlassian.net/browse/RHOSRFE-142 is a downstream tracker i filed to do this "eventualy"
15:58:57 sean-k-mooney but its not someting i have had time to activly work on and i dont expect to actully do that any time soon
15:59:27 winiciusallan[m] i'll be happy to help you with this if it is possible
16:00:02 sean-k-mooney https://etherpad.opendev.org/p/r.bf5f1185e201e31ed8c3adeb45e3cf6d#L747
16:00:09 winiciusallan[m] I need to sync from my side what is the scope for my thesis, there is a slighly chance of testing different strategies...
16:00:36 sean-k-mooney well if you want to talk about this more in the context of watcher feel free to reach out in #openstack-watcher
16:00:54 sean-k-mooney we are also haveing the PTG next week so we could chat about it more there too if you were interested
16:02:11 sean-k-mooney winiciusallan[m]: no worries either way, that the most recent context i can provide on this topic so if you do or do not have scope to wrok on this either is fine
16:03:21 winiciusallan[m] at a first moment, i was thinking of discussing this on nova ptg, but since answered me in advance we can discuss this on watcher's ptg
16:03:31 winiciusallan[m] let's take the discussion to #openstack-watcher
16:03:34 winiciusallan[m] joining there
16:27:38 opendevreview Stephen Finucane proposed openstack/nova master: Fix functional tests and mypy on macOS https://review.opendev.org/c/openstack/nova/+/937727
17:04:05 gmaan bauzas gibi console proxy in threading mode change lgtm now https://review.opendev.org/c/openstack/nova/+/976089
17:26:37 gibi gmaan: ack thanks. Maybe next week before the PTG calls I will have time to get to it
18:15:01 gibi Uggla: sean-k-mooney: I filled out the details of the PCI Grouping topic in the etherpad
18:16:12 sean-k-mooney gibi: ++
18:16:13 gibi and that is my exit today
18:16:15 sean-k-mooney thanks
18:17:54 gibi btw, does AI crawlers know how to read an etherpad?
18:18:16 sean-k-mooney i expoerted it to a file when i wanted it to fix spelling
18:18:21 sean-k-mooney but proably yes
18:18:21 gibi and does our Agentic overlord LLMs can read an etherpad?
18:19:12 sean-k-mooney it can read it in that its just a webpage
18:19:21 sean-k-mooney but i didnt want to givbe it a link
18:19:35 sean-k-mooney because i didnt want it to do random things
19:25:07 Uggla gibi, thanks, but regarding the hour 20:15 looks a bit late and not cool. I wish you a good WE and please rest during the WE. Next week will be intense.
20:43:58 gibi Uggla: thanks. It is complicated. :D
#openstack-nova - 2026-04-19
03:24:52 opendevreview Goutham Pacha Ravi proposed openstack/nova-specs master: Add spec for virtiofs live and cold migration https://review.opendev.org/c/openstack/nova-specs/+/983801
03:26:27 opendevreview Goutham Pacha Ravi proposed openstack/nova-specs master: Add spec for virtiofs live and cold migration https://review.opendev.org/c/openstack/nova-specs/+/983801
04:30:44 opendevreview Goutham Pacha Ravi proposed openstack/nova-specs master: Add spec for virtiofs hot attach and detach https://review.opendev.org/c/openstack/nova-specs/+/983816
#openstack-nova - 2026-04-20
02:19:04 opendevreview Merged openstack/nova master: mem-enc: adjust requirement checks for mem_encryption guests https://review.opendev.org/c/openstack/nova/+/967974
02:19:25 opendevreview Merged openstack/nova master: mem-enc: address minor issues pointed out in the review https://review.opendev.org/c/openstack/nova/+/983278
05:46:50 opendevreview Goutham Pacha Ravi proposed openstack/nova-specs master: Add spec for virtiofs live and cold migration https://review.opendev.org/c/openstack/nova-specs/+/983801
06:16:21 opendevreview Goutham Pacha Ravi proposed openstack/nova-specs master: Add spec for virtiofs live and cold migration https://review.opendev.org/c/openstack/nova-specs/+/983801
06:40:07 tafkamax Hi, I have a question regarding nova-pythonclient. Will commands like nova reset-state <SERVER_ID> be migrated?
06:40:23 tafkamax So when python-openstackclient will become the only one standing.
06:42:37 jkulik Isn't it already available through `opennstack server set --state active`?
06:47:01 tafkamax oh okay, will try
08:47:48 opendevreview Kamil Sambor proposed openstack/nova master: Test nova CLI commands with native threading https://review.opendev.org/c/openstack/nova/+/984036
12:27:28 opendevreview Kamil Sambor proposed openstack/nova master: Enable threading mode for proxy services https://review.opendev.org/c/openstack/nova/+/976089
13:17:48 bauzas as a reminder, the PTG started for now, you can join us by clicking on https://meetpad.opendev.org/nova-2026.2-ptg
14:04:08 dansmith Uggla: didn't you just say we were supposed to be in the cinder room?
14:04:19 dansmith there was some confusion about where to be (here in the cinder room)
14:05:03 Uggla no liberty Nova room (sorry if it was not clear)
15:15:04 melwitt gibi: thanks for checking on the nova-vtpm patch. I saw it failed again :( so I will investigate
15:36:01 gibi Uggla: your nova bug stats will not like my effort of filing bugs for eventlet tech debt https://bugs.launchpad.net/nova/+bugs?field.tag=eventlet-removal
15:37:10 gibi melwitt: ack, thanks for trying to stabilize it
15:42:00 stephenfin tafkamax: https://docs.openstack.org/python-openstackclient/latest/cli/decoder.html
15:56:18 opendevreview Merged openstack/os-vif master: Bump min version of pyroute2 https://review.opendev.org/c/openstack/os-vif/+/984934
16:10:10 opendevreview Stephen Finucane proposed openstack/nova master: objects: Prepare for oslo.versionedobjects 3.10.0 https://review.opendev.org/c/openstack/nova/+/985506
16:10:30 tkajinam I'm just wondering how memfd works for allocation. Should we still calculate memory allocation according to the numa topology ?
16:10:48 tkajinam (assuming we still have to align cpu amd mem in numa-aware way, looking at it may be used for ovs-dpdk
16:11:26 tkajinam by the way memfd is heavily used in cc case (actually instances with snp support or tdx support uses it implicitly and it's interesting after learning the mechanism quite recently
16:22:28 Uggla @gibi, I think the bug you opened are valid so can be triaged. Anyway that's not too bad.
16:23:21 Uggla tkajinam, good to know memfd is something interesting for CC case.
16:23:40 Uggla tkajinam, if you have a doc pointer on it I'm interesting.
16:24:10 Uggla tkajinam, probably a good point to convince my mgmt that this feature is interesting.
16:46:03 tkajinam Uggla, let me find one, but it might not match your use case largely because 1) memfd is implicitly used by qemu 2) the main objective of using memfd is not performance but allocating private/encrypted memory for vm
16:50:00 Uggla tkajinam maybe part of an answer to your question --> https://lwn.net/Articles/990663/
16:51:05 tkajinam ok so you still need to maintain memory:numa mapping (seeing that host-nodes=0)
16:52:18 tkajinam Uggla, this might be too-much information but captures the qemu's internal logic to leverage memfd for private memory area https://review.opendev.org/c/openstack/nova-specs/+/983376/comment/658b8a62_20071ca8/
16:54:29 sean-k-mooney tkajinam: memfd can have the shared flag set or not i bleive
16:54:41 tkajinam yeah
16:54:46 sean-k-mooney i woudl prefer to always have memfd shared=true
16:55:13 sean-k-mooney but keep "annoumus" spelled correctly for those that want to opt out
16:55:27 sean-k-mooney i.e. evnetually memfd i think shoudl become the default
16:55:38 sean-k-mooney but ya in newer qemu memfd does not need numa
16:55:44 sean-k-mooney but it did when it was first done
16:57:24 sean-k-mooney the fact that memfd does not require numa affinity is one of the reaosn its intersting

Earlier   Later