Subject Re: [firebird-support] Re: connection problems with RedHat v. 9
Author William L. Thomson Jr.
Having read through all threads I think you all are quick to blame the
OS. When the problem very possible may lie else where like the env,
config, or client program design.

There has been no mention of the isc_config file or it's contents in
relation the the env.

What type of system and resources are available to Firebird?

Are you monitoring the server's load when connections are dropped and/or
refused? What's the proc and memory usage?

It seems from a previous post that something is causing the server to
crash and be restarted?

What are the clients doing at this point?

Have you checked all logs, not just Firebirds?

On Sat, 2004-01-24 at 06:13, Peppe Polpo wrote:
> <gkrishna@j...> wrote:
> >Windows 2000 continues to work fine and performance is improved
> (working satisfactorily - at least 3 to 6 months). Linux once a week
> or once in 10 days, the client when doing multithread access gets
> 10054 error and loses the connection.

What type of client, Linux or Windows or something else?

It's rare for the server to create it's own problems. Almost always the
cause comes from something the client did. Or a env that is not properly
configured when the client is doing something intense?

> what a shame! Windows works and Linux does not. Bill would pay to
> hear this :)

What facts is that assumption based on? You would have to have identical
machines, from a hardware stand point of view. Firebird would have to be
configured to use similar resources on both from a software stand point,
and etc.

There is much more to configuring and dialing in a Linux machine than
Windows. Linux is more custom fit, Windows is more universal.

However the time spent up front with Linux is made up in the long run
with reliability and stability.

> I found several threads in NGs reguarding connection problems with
> Interbase on a RH server, but I cannot understant to what extent they
> reguard FB too.

Notice those are with InterBase not firebird. Meaning those problems
most likely are in the past. Common ones that Linux users ran into and
no longer run into do to those threads existence. I bet I read one or
two of them when first installing InterBase on RH 6.2 in 2000.

> I also found out that IB is not certified on RH v. 9 (see

? It's not like you are running an enterprise version of RH. Bye the way
are you using InterBase or Firebird? Wouldn't a better comment be
Firebird is not certified?

When using a regular Linux distro why would a cert matter? Most of the
programs and the env is not certified against it each other. Things that
regardless if Firebird/IB was certified, what it depends on may not be.

So really how can you certify the app. That's pretty much the bases
behind the enterprise editions of RH. They are taking the time to go
through everything for stability and reliability. Which obviously takes
time and resources, thus you paying to for a license to use the
Enterprise versions.

> Unfortunately I trusted some messages in this NG stating that ev was
> fine with FB and RH9.

I have run Firebird on RH 6.2/7.0/7.1/7.2/7.3/8.0/9 and now Gentoo
without any problems in all types of different env and uses.

With Linux and Windows clients. No mac or alternative clients yet.

The errors you see in your logs are common, except the one with error
code 32. I have not seen that one in my logs. Then again my Firebird
server is not crashing?

> I am trying to isolate the problem, for instance now I am not using
> But if the problem persists, I will have to go back to an older RH
> version, or to a different Linux distro.

Regardless of distro I think your problem lies else where possible in
how the machine was configured. Or how the client program is designed.

I would say that Windows has a higher tolerance for errors than Linux.
Thus a quick and dirty app running fine on Windows and having to be
adjusted and fine tuned to run on Linux.

Also just out of curiosity, have you tried CS on Linux?

William L. Thomson Jr.
Support Group
Obsidian-Studios, Inc.