[erlang-questions] Death of Erlang VM

Per Hedeland per@REDACTED
Fri Sep 19 08:54:22 CEST 2008


"Brian Troutwine" <goofyheadedpunk@REDACTED> wrote:
>
>I've encountered a problem, but I'm not really sure what is the
>matter. After running my erlang application for some time I was
>noticed that it had died, though not exited or become a zombie.
[snip]
>Sep 18 17:33:05 valinor kernel: [1134774.645453] Oops: 0000 [1] SMP

The "cute" 'Oops' is telling you that the Linux kernel has discovered
some badness in its internal data structures - this is almost always
indicative of a kernel bug. It may or may not have anything to do with
the fact that you were running Erlang, but one aspect of the Linux
'Oops' handling is that the currently running process is logged and
killed, regardless. (Another is that the kernel keeps running, which
might be considered an odd thing for the entity responsible for system
integrity to do in that situation - a traditional Unix kernel forces a
reboot, with a crash dump if enabled.)

Anyway, even if 'beam' somehow triggered it (and a quick look at the
stack trace indicates that this is likely), it's still a kernel problem
- by definition, an unprivileged user process should never be able to
mess up the kernel data (normally a privileged process shouldn't be able
to either, but if it can and does scribble on kernel memory via /dev/mem
or somesuch, it may be hard to argue the case).

If not a kernel bug, it can of course be a HW issue - e.g. faulty memory
or unreliable power supply can cause arbitrarily weird problems. Your
best bet is probably to take the issue to some Linux support forum.

--Per Hedeland



More information about the erlang-questions mailing list