When a virtual machine is created, the corresponding virtbmc pod is created but unable to transition to a ready state. Instead, it keeps crashing with the following error message:
time="2024-11-11T03:15:04Z" level=info msg="Starting virtbmc"
time="2024-11-11T03:15:04Z" level=info msg="Initializing the simulator..."
time="2024-11-11T03:15:04Z" level=fatal msg="unable to run the ipmi simulator"
Note that the exact error was not exposed in the original main branch codebase, so I made a couple of changes to let it show the hidden error message. It turned out it was a bind: permission denied error.
To Reproduce
Steps to reproduce the behavior:
Create a Kind cluster
Apply KubeVirt's VirtualMachine CRD
Deploy cert-manager
Deploy KubeVirtBMC
Create a dummy VirtualMachine object
Observe the virtbmc agent pod crashing with the error message bind: permission denied
Expected behavior
The agent can bind the network socket and works normally.
Screenshots
Environment (please complete the following information):
OS: Kind v1.28.13
KubeVirt version: v1.1.1
KubeVirtBMC version: main
Additional context
This was first found when I was developing the e2e testing. My environment was on Apple Silicon macOS but it can also be reproduced in a linux/amd64 virtual machine.
However, the same issue does not seem to be happening on Harvester clusters. I can't find the difference that caused the issue.
Describe the bug
When a virtual machine is created, the corresponding
virtbmc
pod is created but unable to transition to a ready state. Instead, it keeps crashing with the following error message:Note that the exact error was not exposed in the original main branch codebase, so I made a couple of changes to let it show the hidden error message. It turned out it was a
bind: permission denied
error.To Reproduce Steps to reproduce the behavior:
virtbmc
agent pod crashing with the error messagebind: permission denied
Expected behavior
The agent can bind the network socket and works normally.
Screenshots
Environment (please complete the following information):
Additional context
This was first found when I was developing the e2e testing. My environment was on Apple Silicon macOS but it can also be reproduced in a linux/amd64 virtual machine.
However, the same issue does not seem to be happening on Harvester clusters. I can't find the difference that caused the issue.