Fri Jun 17 07:48:42 CEST 2005
That doesn't sound good, can you email me your test, and I'll have a look
Mickael Remond writes:
> I have encountered potentially strange behaviour with Mnesia, when the
> changes applied to the database where made in a script-like fashion in a
> short-lived program.
> I had a program that was starting the database, then was inserting a
> record in the database and last was stopping mnesia and the Erlang VM with
> My record where not found in the database after a restart of the VM and
> Mnesia. It where correctly written by my script however (I put a read and
> print instruction to check).
> I found that the record was in the log file in the database directory, but
> was not replayed by Mnesia when restarted.
> I thought that it might have to do with log threshold options (Dump log is
> not triggered as the size of the transaction or the max time before dump
> is not reached).
> Finally adding a mnesia:dump_log instruction before stopping Mnesia in my
> short lived script solve the problem.
> I was wondering if this was the expected behaviour. I would have guessed
> that Mnesia should have read the log on startup or that shutdown would
> have dump the log.
> I was using Erlang/OTP R10B-3.
> Thank you for your feedback
> Mickaël Rémond
More information about the erlang-questions