Hi All,

I have a debian64 running FB 2.5. It serves 19 databases. Several 
clients connect to this server, amongst them Lazarus services, PHP and 
remote Delphi connections.

About once a week, the system locks up. Not just connections to one 
database, *all* connections to any database freeze. New connections are 
impossible.

I have to kill fb_smp_server.
It's immediately restarted by fbguard, and then everything works well again.

Server cpu usage is only about 2% during the lockup, there are no traces 
in the logs of anything unusual.

I know I have to review the transaction handling in my code, but still 
I have one question:

If a lock occurs in a database, is it normal behaviour that *all* 
connections from *all* clients to *all* databases lock up?  I thought 
that only the database where the lock occurs would be affected, or even 
only connections to the same table, or even transactions that touch the 
same records in a table. But this doesn't seem to be the case, 
everything freezes.

Any thoughts?

Thanks,
John


Reply via email to