Sean asked that this run through firebird-support first, so here it is:

I found my (2.1.3 classic on windows server 2008) staging server frozen this
morning -- processes running, but all clients hung. firebird.log shows the
following (I hadn't paid any attention to it on friday, didn't notice that it
was down over the weekend.)

BTFIPS Thu Apr 29 17:47:01 2010
operating system directive CreateFile failed

BTFIPS Thu Apr 29 17:47:01 2010
The requested operation cannot be performed on a file with a user-mapped
section open.

BTFIPS Thu Apr 29 17:48:52 2010
operating system directive WaitForSingleObject failed

BTFIPS Thu Apr 29 17:48:52 2010
Cannot create a file when that file already exists.

... then repeat last two errors forever (about every minute, may correspond to
cron job trying to connect.)

Not a PDB build, so I don't have anything else to give you. Drives (OS &
storage) have hundreds of gigs of space available, at least currently. The
messages don't indicate what file was being attempted, so that leaves it up to
guessing.

Note that the box had no CPU usage at the time (it wasn't crunching on any
large queries), and further attempts to connect to FB would just hang. There
were maybe 8 or so FB processes running when I found it.

The next scheduled nbackup would have been for 18:00 that night; the 17:00 run
finished successfully. There's a cron job that checks every minute or so for
any long-running statements via the mon$ tables (it kills them only if the SQL
begins with a comment block stating it should be killed if it runs too long.)
Apart from that, there would have been no usage on this server at the time.
Tomcat maintains a small connection pool, which would have been idle.

According to the log file, it looks like this has happened before -- in
november of last year. So it's rare, at the very least, and fixed by restarting
firebird. I didn't see any matching cases of this in past firebird-support
messages about freezes.

-Philip