I'm running FB1.5 SS on WinNT server. I have seen a few occasions, one of
them being this morning, where the FBServer.exe runs at 100% of processor
time. It will allow connections, but those connections can do nothing.

This morning, I have stopped all connections to the database. I let it run
for two hours, hoping that whatever was going on would finish. During this
time, I was able to run gstat, and found the difference between OAT and
next transaction is only 6. Sweep interval is set to 0.

After two hours, I checked the Performance monitor for disk reads and
writes, and found no activity. I then tried to stop the FB service, which
hung for about 10 minutes with the 'Attempting to stop Firebird Server' ,
but the process was still running at 100%. I then rebooted the machines.

I have heard comments when this question was asked before of 'are you sure
the sweep thread isn't running?'. My question is, how do I know if the
sweep thread is running? What can be done in these situations to bring the
server back into a more responsive mode, short of rebooting the server?

The only other strange thing that happened between last night (when it was
working correctly) and this morning (when it was not) was a backup job that
is run from another server failed. We backup from the main server via a
batch file called from a second server and store the backup file on that
second server due to space constraints. My backup logs show the backup
procedure starting, with an error about not being able to open the backup
file (we copy over the backup file each night). One of the technicians had
tried to copy the backup file to an external driver earlier yesterday, and
it seems that action left an open lock on the backup file. This couldn't
have anything to do with the 100% processing, could it?

tia,

Bob M..