Subject Re: INET/inet_error: read errno = 9 messages + Server becomes unresponsive

I think I have found solution to this problem for our specific case.

It seems to be due to the use of multiple CPU's.

From the following article it seems that Super Server (any version) does not support this at all!:

Apparently the solution is to use Classic server, but again this does not seem to work entirely well according to the article, and with a high number of users connecting it seems this could cause the same problems.

So, i have disabled all the other CPU cores apart from one, and now in the firebird log the INET error=9 messages have stopped completely!! Considering that we were getting one of these messages roughly every second this does seem to have been the problem.
Also for the past 10 days, we have not experienced a server crash.

Since we are running on Linux the CPU's have to be disabled at kernel level (editing the CPUAffinity setting in firebird.conf only works on Windows).

To show how many processors are currently online run the following command:

cat /proc/cpuinfo

To disable a CPU core:

Hopefully this may help in your situation....

For now, it seems we will have to run in this configuration until the new Firebird 3 comes out with proper support for multiple CPU's, or switch to Classic, but again run with a smaller number of CPU's.

--- In, "Mark Rotteveel" <Avalanche1979@...> wrote:
> Since error 9 is 'bad file descriptor' on Linux, you might want to check if your system has sufficient file descriptors available.
> Mark
> > As I said, this problem is very similar to the one I posted in the thread
> > "Repeated database corruption". There I explain about the cases when the
> > database got corrupted and I have to repair it using gfix -mend followed by a
> > backup and resotre. But there are many case when stopping and restarting
> > FireBird service solves the problem. In some cases this does not solve the
> > problem and I have to restart the server. And in some cases I try to stop
> > FireBird service and it never stops and so I have to restart the server. If
> > after this, I check the database with gfix -v -full, in some cases it is
> > perfect but in other cases I need to repair the database.
> >
> > In this network I used firebird 1.5 SS more than a year without any
> > problem. Then I changed to SS and it worked some months without any
> > problem but in 04/2009 the problems began. In 06/2009 I changed to
> > SS and the problem continues. In these last days I changed to
> > CS but it did not solve the problem.
> >
> > As you say, it seems to be a very 'mysterious' problem. I could not find
> > any relevant change made to the hardware or software by the time the problem
> > began. Perhaps is a combination of many things.
> --
> GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT!
> Jetzt freischalten unter