At 01:37 PM 1/03/2005 +0200, you wrote:

>my firebird server hangs, I don't know thre reason because nothing is
>reported in logs, I can give you ps output and tell that I am using
>rfunc library and there are 30 connections every 5 seconds. Maybe
>database needs some network tweaking? Iuse superserver version 1.5.1. It
>hangs randomly one time per 12h approx.
>
>ps afx output(see Rl state):
>24209 ? S 0:00 /usr/lib/firebird2/bin/fbguard -f
>24211 ? Rl 26:48 \_ /usr/lib/firebird2/bin/fbserver

If this is ps -afx output, then no clients are connected. R means it's
using CPU. What does "I" signify - is this something peculiarly Debian?

What does a "hang" look like to you? i.e. describe what happens (or
doesn't happen) when you think it is hanging.

I'm tempted to suggest that you are seeing a GC or sweep thread running,
though the ps tree output should show it.

>What do you think about rfunc UDF library?

Can't answer that - I don't use it.

>btw, rfunc is using ib_util_malloc function from libib_util.so from
>firebird superserver 1.5.2 package which I installed separetly(I
>couldn't find debian package containing this file), server is installed
>from debian package.

I didn't know of any changes to ib_malloc between subreleases. I think
we'd have heard a lot of howls from UDF users if it was a problem.

>What could you suggest?

Look at limits.conf to see whether there is a cap on the number of TCP/IP
clients that can be connected? A shot of ps -afx | grep fb taken during
one of these hang-ups would be interesting.

./hb