Replies: 1 comment
-
Need to investigate how crash dumps generated in eBPF binaries in production can be consumed for investigation. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Describe the bug
We currently have 'asserts' for un-recoverable and/or fatal errors. These asserts are triggered in debug builds but are compiled out in 'release' builds.
This can often lead to silent failures and/or buggy system behavior in release builds. Also, considering that most ebpf consumers would prefer running the retail builds (ebpf program verification times are un-acceptably slow in debug builds), the subsequent inconsistent behavior would only end up confusing the ebpf end customers.
We should bugcheck the system and generate a crash dump to debug and fix issues such as:
OS information
No response
Steps taken to reproduce bug
N/A
Expected behavior
The system should bugcheck and generate a crash dump.
Actual outcome
The failures are ignored and system behavior becomes indeterminate.
Additional details
No response
Beta Was this translation helpful? Give feedback.
All reactions