corrupt external term

[Brian Acton]

Hi everyone,

I'm running R13B04 on FreeBSD 7.3-stable. I have a heterogeneous cluster set
up in a frontend/backend configuration with two mnesia nodes doing primary
storage in the backend.

The frontend nodes talk to the backend machines using the mnesia remote
protocol.

Recently, I've started to experience failure as seen by this

'frontend-node1' got a corrupted external term from 'backend-node1' on
distribution channel 10173
<<...,104,2,114,0,3,82,0,3, ...>>
ATOM_CACHE_REF translations: 0='frontend-node1', 1='', 2=xmlcdata,
3=xmlelement, 4=jid, 5=never, 6=offline_msg

The binary above is truncated for brevity but I think everyone gets the gist
of what is going on.

A quick google search on the phrase yielded only one link. This link
references a bug in the atom cache:
http://groups.google.com/group/erlang-programming/browse_thread/thread/0d5bf2cbaa7c921a/5b59ee7d0800cde7?lnk=raot&fwc=2&pli=1

So several questions ensue:
1) is this just a simple matter of mnesia table corruption?
2) any recommendations on how to rebuild / resolve the corruption?
3) is it possible that this corruption is related to the aforementioned atom
cache bug?
4) any tips on instrumentation / tracing / logging to further isolate this?

Any tips / tricks / help, definitely appreciated.

Thx,

--b