Hey @Christian, happy new year!
Test setup and results so far.
- You shared 4 versions for the bisect:
- v10.0 bug confirmed (unhealthy)
- v10.2 unknown
- v10.4 unknown
- v10.6 bug not present (healthy)
- I only had 3 KVM hosts available for testing, so for the first round I used:
- v10.0 as the control sample
- v10.2 unknown
- v10.4 unknown
- Results from the first test showed that all three versions had the CPU
climbing issue.
- For the second round, I needed to compare against a healthy version, so I
switched to:
- v10.2 unknown
- v10.4 unknown
- v10.6 as the control sample
Note: I created an automation that for 3 weeks deployed and destroyed
200,000+ VMs to generate load and trigger the issue. Unfortunately, it
caused the environment to become unstable and the second test failed.
Looking back, my load test was not very efficient at triggering the
issue. Our servers are large (40 cores, 256 GB RAM each), so they are
hard to push to their limits.
Today, I changed the load-test approach to try different tasks that may
trigger the condition with less raw load and more diverse behavior.
By next week, more hosts should be joining the cluster, and I hope to
share more meaningful and useful information then, rather than just a
story 😊
Regards,
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2133531
Title:
libvirt 10.0.0-2ubuntu8.* from 24.04 misbehaves affecting CloudStack
Agent
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/2133531/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
