containerd / containerd/containerd
Make it POSSIBLE to debug cdi hooks
- Dominant language
- Go
- Stars
- 21.3k
- Forks
- 4.1k
- Avg merge
- 2d 4h
- Merged PRs (30d)
- 92
Description
### What is the problem you're trying to solve
Not sure if this is something that needs to be resolved here or in the runc implementation, but its ABSOLUTELY INSANE that all you can get from logs upon a cdi failure is this:
```
time="2024-12-30T19:09:23.702691687-07:00" level=error msg="StartContainer for \"a596026fcd53b01f175a6960734548967eca8687fbdf5b4bc310fe2f25c6d5cb\" failed" error="failed to create containerd task: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: signal: illegal instruction (core dumped), stdout: , stderr:: unknown"
```
### Describe the solution you'd like
There needs to be a way to debug what exactly it is that is going wrong with the hook.
Ideally, it should be possible to see logs from the hook command. strace information in the event of a core dump or some type.
Also, when trace logging is enabled, the exact call being made to runc that is being made should appear in the logs.
### Additional context
_No response_
Contributor guide
Assessment
This issue has not been assessed yet.