John B. Moore wrote:

> In a previous thread I discussed a problem I was having with Firebird
> on a specific machine.. I have now been tracking the behaviour of
> Firebird on the problem machine vs a machine that is working
> correctly. Both have the exact same OS & version, and Firebird &
> version. (CentOS 4.2 and latest Firebird SServer downloaded Sunday)
> and nearly identical hardware.
>
> On the Problem machine, when checking "netstat" and looking at the
> "gds_db" service connections I see mostly CLOSE_WAIT states.

I see that nobody comments, so I'll try to give you some hints. I did
see a similar behavior on one Debian Linux box, and I believe it was
NTPL issue (using wrong SS version for the kernel).

Does "nearly itdentical hardware" mean that CPUs are the same? Do you
prehaps have a CPU with HT and possibly enabled on one and disabled on
other system?

What about kernel versions? I assume they are the same?

What about clients that access the Firebird server? Do they all have the
same version of Firebird client library? Do they access both of these
servers?

In the end, you might try switching to ClassicServer. It solved the
problem in my case, and I didn't have the time to trace the problem to
the end (customer's site far away - no network access).

You should also let us know which is the "latest Firebird SServer"
package you downloaded... 1.5.3? 2.0?

> On the Machine that is working (and currently has all of the databases
> that were on the original problem machine), ALL connnections remain
> "ESTABLISHED" (observed over the last 24 hours)

Now you confused me. Is the CLOSE_WAIT problem showing up _after_ you
close the connection or _during_ the connection?

--
Milan Babuskov
http://www.flamerobin.org