[erlang-questions] Mnesia ETS crash

[Mike Oxford]

Further information:
1)  The table is question does exist, has existed, and has been running for
a while.
2)  safe_fixtable fails with badarg if the table does not exist.

#1 runs for a good long while (month?) and then suddenly #2 pops up.    Why
does mnesia 'lose' the table?

Here's more context

<snip>
===== ALIVE Tue Jul 31 17:22:07 PDT 2012

===== ALIVE Tue Jul 31 17:37:07 PDT 2012

===== ALIVE Tue Jul 31 17:52:07 PDT 2012

===== Tue Jul 31 18:04:17 PDT 2012

=ERROR REPORT==== 31-Jul-2012::18:04:17 ===
Error in process <0.15759.10> on node '<redacted>' with exit value: {badarg,
[{ets,safe_fixtable,[px_channel_log,false]},{mnesia_lib,db_fixtable,3},
{mnesia_loader,cleanup_tab_copier,3},{mnesia_loader,send_table,3},{mnesia_controller,send_and_reply,2}]}


=====
===== LOGGING STARTED Tue Jul 31 18:04:19 PDT 2012
=====
Exec: /usr/local/chat-game-gwm/current/erts-5.8.4/bin/erlexec -boot
/usr/local/chat-game-gwm/current/releases/1/chat -mode
embedded -config /usr/local/chat-game-gwm/current/etc/app.config -args_file
/usr/local/chat-game-gwm/current/etc/vm.args --
 console
Root: /usr/local/chat-game-gwm/current
Erlang R14B03 (erts-5.8.4) [source] [64-bit] [smp:24:24] [rq:24]
[async-threads:64] [hipe] [kernel-poll:true]


=ERROR REPORT==== 31-Jul-2012::18:04:22 ===
Mnesia('gwm_game@<redacted>'): ** ERROR ** mnesia_event got
{inconsistent_database, starting_partitioned_network,
'gwm_game@<redacted>'}
DEBUG: mnesia_helpers:118 - "{ \"Connecting to cluster: \" , NodeList }"

</end_snip>


On Mon, Aug 6, 2012 at 5:04 PM, Mike Oxford <moxford@REDACTED> wrote:

> =ERROR REPORT==== 18-Jun-2012::15:01:49 ===
> Error in process <0.1743.0> on node '<redacted>' with exit value:
> {badarg,[{ets,safe_fixtable,[px_channel_log,false]},{mnesia_lib,db_fixtable,3},{mnesia_loader,cleanup_tab_copier,3},{mnesia_loader,send_table,3},{mnesia_controller,send_and_re
> ply,2}]}
>
> Logging shows it was then restarted and {inconsistent_database,
> starting_partitioned_network...
>
> mnesia is Erlang R14B03 (erts-5.8.4) [source] [64-bit] [smp:24:24] [rq:24]
> [async-threads:64] [hipe] [kernel-poll:true] so it shouldn't fall victim to
> the safe_fixtable/next issues from R13x
>
> This has happened to us twice now, both times exactly the same error.
>  System runs other ram_only nodes as well with no issues; this one is a
> disc_ table with light load.
>
> Thoughts?
>
> -mox
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20120807/2e662e8f/attachment.htm>