On 24-2-2013 00:43, jackmaso...@rocketmail.com wrote:
> We converted from Windows 7 Interbase 6.02 last Monday by backing up the 
> databases then restoring them to SuperClassic under Mint. The hardware is an 
> Intel I3 with 8GB of memory, a hardware RAID (2 SSDs), and a system SSD.  All 
> 5 Firebird databases are on the RAID, a shared drive with Windows 7.
>
> All 6 users are on Windows 7.  Everything seemed to be running fine until 
> this afternoon.  The first indication was a temporary server disconnect for 
> one user.  We restarted his program and all seemed well.  About an hour 
> later, all users failed, unable to write to the database. In over 12 years of 
> using Interbase 6 on Windows, we have not encountered this problem.

What were the specific errors received?

> When we rebooted the Firebird server, everything returned to normal 
> operation. At the time, we did not know the correct name for the Firebird 
> process so we could not tell if it was still running before the reboot. We do 
> not know where the Firebird logs are located, so cannot check them to see 
> what went wrong.

It sounds like this transition wasn't really tested (both from the 
perspective of the application and operations).

> Should we be using a different Firebird server for this setup?  Can we 
> determine the cause of the failure after having rebooted?  We were having to 
> reboot the Windows 7 Interbase server at least once a week and do not want to 
> have to return to that scenario.  We expected Firebird to run for many months 
> without a reboot.

Most likely with Interbase 6.02 you were using SuperServer, which shares 
a single page cache for all connections to the same database. Now with 
SuperClassic each connection has its own page cache. If you used to have 
an extremely large page cache configured then you could simply be 
running out of memory.

Also a lot has changed between Interbase and Firebird 2.5, maybe your 
application is doing things 'wrong' (== incompatible with Firebird 2.5), 
but without specific errors that is impossible to tell.

Mark
-- 
Mark Rotteveel

Reply via email to