Skip to content

Conversation

@tsarajar
Copy link
Contributor

@tsarajar tsarajar commented Jun 2, 2025

Description

When looping through VMs in the scheduler, not all previous data was cleared before or at get_capacity. This leaked old VM requirements to the next one blocking the scheduler in certain circumstances. This fix will ensure a clean struct for each VM in the loop.

Branches to which this PR applies

  • master
  • one-X.X

  • Check this if this PR should not be squashed

@tsarajar tsarajar changed the base branch from master to one-6.10-ce June 2, 2025 06:59
@tsarajar tsarajar mentioned this pull request Jun 2, 2025
3 tasks
In scheduling loop HostShareCapacity object contained numa
nodes from previous virtual machine. This caused scheduling to fail
sometimes even if there would be a suiteable host. Fixed by
recreating the object inside the loop for every virtual machine.

Signed-off-by: Tony Sarajärvi <tony.sarajarvi@qt.io>
@paczerny
Copy link
Member

paczerny commented Jun 8, 2025

It's already fixed in master (version 7.0). Apply only to 6.10 branch

@rsmontero
Copy link
Member

OK THANKS! ❤️

Merged in 6.10 branch for 6.10.5 release

@rsmontero rsmontero closed this Jun 16, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants