Subject RE: [IBO] Linux server dropout
Author Jason Wharton
> Lester Caine wrote:
> > Lester Caine wrote:
> >>>> Anybody got any ideas why I should be getting problems
> with this running into
> >>>> Linux? Any suggestions what I need to look at to find
> out how to fix it! One
> >>>> thought was the time-out on the connection, but why
> would that be a problem.
> >>>> The Linux server is a fast AMD based 64 bit machine and
> we have been finding a
> >>>> few niggles with this set-up and PHP, but it should be
> stable on legacy
> >>>> connections?
> >>> OK Just dropped out again - and I have three INET-104
> errors in the Firebird
> >>> log. I'm just not sure why the program is bombing if it
> miss reads :(
> >> No messages in the log following the next drop-out.
> >> I have a MainSessionError set up which should be giving an
> error message
> >> pop-up before closing the program, but this seems to be
> bypassed by what ever
> >> is causing the crash :(
> >
> > I've added some extra logging of connection stuff, but
> since killing the
> > sophos anti-virus the machine has run perfectly today. I'll
> give it a few more
> > days but either the anti-virus was messing with traffic, or
> the extra checks
> > I've added have changed something. To prove a point I'll
> leave alone until mid
> > week and then drop back to the previous build to see if
> that is also stable.
> OK I'm not out of the woods here :(
> All that killing the anti-virus crap has done is allowed the
> machine to run a
> lot faster, so that a restart is completed faster. While this is only
> happening three or four times a day, it's making IBObjects
> unusable with Linux
> servers. I've had to switch back to Windows for next week to
> keep the customer
> happy, but the path forward means that I MUST sort this out.
> Even the extra logging added to the BCB6 program has not been
> fired when the
> program closes. There is nothing in the Firebird log that
> reliably corresponds
> with the crash. Everything I throw at the program to try and
> create the
> problem simply gives a 'clean' error report. The same program
> is running on a
> number of windows powered sites without any problem, and I
> expect to see no
> restarts in the coming week. It's only one of the programs
> connected to
> Firebird at the time and the others carry on fine -
> restarting the failed
> program when it is found to have stopped on the next minute.
> While I can
> modify checking to ensure a faster restart after a crash, the
> ticket printer
> then throws out a '9999' test ticket which it would be
> difficult to prevent.
> ANY ideas what I can try next ????
> Jason please - what on the database connection side could
> cause an immediate
> failure.


I've thought about this some and the only thing I can think of is perhaps its picking up the wrong client library. Have you tried to make sure the correct client library is being used?