General protection error in beam.smp

[erlmachinedev]

Describe the bug
The Beam experienced crash during RabbitMQ execution. The Beam is running as containerized application within k8s cluster.

To Reproduce
Despite the high complexity and internal nature of the issue (accidental runtime segfault) there is no way to reproduce. The issue itself isn't deterministic.

Expected behavior
The Beam is operational and no Pod restart happens.

Affected versions
OTP-26.2.5.3

Additional context

Check SNMP counters.[14470870.648422] traps: erts_sched_4[22785] general protection ip:5615da3dea02 sp:7fd72c01f510 error:0 in beam.smp[5615da1aa000+309000]

This Log snippet indicates OTP-26.2.5.3 crash followed the Pod restart. We classified the issue as a part of Erlang runtime rather then RabbitMQ or whatever app is running on a top.

The information attached is minimal (only Log snippet as additional context below) but we assume that maintainers:

• correlate this issue with similar one or well known preceded
• suggest the steps to reproduce the issue
• request additional information which helps triage the issue

Thanks for you maintaining effort and please suggest the next steps,

Best, Slava

[garazdawi]

correlate this issue with similar one or well known preceded

OTP-26.2.5.4 has fixes for segfault.

suggest the steps to reproduce the issue

You can try to run the debug emulator if you have a testcase/simulation that you think should trigger the error.

request additional information which helps triage the issue

We need to have a core dump from the fault together with a non-stripped beam.smp binary. Without that and no way to reproduce there is not much that we can do.

[garazdawi]

Closing due to inactivity. Feel free to open again.