Well, its seems you have certainly tried a few options to cure this problem.

This has only really started to be a problem for us on one of our servers after we moved the database to a new machine a few weeks ago. We were previously running the database on a dell poweredge 1850, with the same firebird version (2.1.2.18118 SS), and it had been pretty much rock solid for 6 months. We only moved to the Dell R710 due to it having multiple disks for failover.

We also upgraded another server that was running firebird 1.5.3 to 2.1.2.18118 and also new hardware (dell R610 + debian 5), and this does have the 'INET/inet_error: read errno = 9 messages' but not as often, and as of yet has not crashed (been up for 50 days with no issues).

We also have another server running Redhat 9 with a 1.5.3 SS version of the firebird database, and that also has the INET errors, and has been having some problems with random crashing (i.e. firebird process still running, but not allowing connections) but after a restart is OK again. The last occurence of this was on Oct 12th, but it has been running fine since then.

So, this problem is very 'mysterious' in that it does not seem to follow a logical pattern, making it hard to diagnose where the actual problem is! i.e. it does not seem to be a specific hardware, DB version, OS problem....

How do you connect to firebird, is this through a client? We are connecting via a PHP application, not sure if this does have any relation to the problem though.

Is your database running any better on Windows 2008? What version of Linux were you running?

--- In firebird-support@yahoogroups.com, "andreas_kc" <andreas_kc@...> wrote:
>
> I have experienced a number of issues just like this in the past 10 months. I wish I could help, but I don't know the answer. I have tried a bazillion different solutions. I've ripped out switches, changed the server hardware, changed cables, tested NIC cards, and prayed to the voodoo gods. I usually run Firebird on a Linux server as well, but today is the first day we are running our production database on a Windows 2008 Server.
>
> I'm crossing my fingers that this behaves better on W2k8.
>
> --- In firebird-support@yahoogroups.com, "jay_r42@" <jay_r42@> wrote:
> >
> > Hi,
> >
> > We have recently installed firebird 2.1.2.18118 Super Server onto a Dell R710 running Debian 5 (lenny).
> >
> > The firebird server appears to run fine for a while (anywhere between 1-5 days) but will then become unresponsive (i.e cannot connect to the database). The firebird process is still running, so the only way to resolve has been to restart the firebird service and connections to the database can then resume.
> >
> > In the firebird.log i have noticed frequent messages regarding 'read erno = 9' shown below.
> > I have checked the FAQ on the firebird website (http://www.firebirdfaq.org/faq120/) for the meaning of this message and it seems to be related to a bad file number?
> >
> > We have two other servers running firebird and while they do get these errors it is not as frequent.
> >
> > If anyone could offer advice on how to stop these messages appearing, and also clarify if this is the reason the server becomes unresponsive it would be most apreciated.
> >
> > Thankyou,
> > Jason
> >
> > mercury2 (Server) Thu Oct 22 15:50:33 2009
> > INET/inet_error: read errno = 9
> >
> >
> > mercury2 (Server) Thu Oct 22 15:50:33 2009
> > INET/inet_error: read errno = 9
> >
> >
> > mercury2 (Server) Thu Oct 22 15:50:34 2009
> > INET/inet_error: select in packet_receive errno = 9
> >
> >
> > mercury2 (Server) Thu Oct 22 15:50:37 2009
> > INET/inet_error: read errno = 9
> >
> >
> > mercury2 (Server) Thu Oct 22 15:50:38 2009
> > INET/inet_error: select in packet_receive errno = 9
> >
> >
> > mercury2 (Server) Thu Oct 22 15:50:40 2009
> > INET/inet_error: read errno = 9
> >
>