Apache buffering error log warnings

Hello,
I have found ran into a problem for which I am now guessing the
cause... but I need some help.

I have two Apache Servers running with pretty similar conf (modperl
scripts). They both output a large HTML file.
Server #1 run and outputs the data fine but produces many warning on
the error log.
Server #2 freezes somewhere on the way. Since output warnings to the
error log are buffered (I know this from just watching the error log
with tail). I guess that the whole scripts freezes because the buffer
gets full, right?
So, first of all, can someone confirm this?
Second, do I have any control over the size of the buffer? One server
works, the other does not, so I guess the buffers are different size?
Otherwise, can someone point out what other configuration variables
could affect to identical scripts running in almost identical
environments.

Thank You.

On 2 Aug 2004 05:17:38 -0700, jaime1974@yahoo.com (SuperPoofie) wrote:

>Hello,
>I have found ran into a problem for which I am now guessing the
>cause... but I need some help.
>
>I have two Apache Servers running with pretty similar conf (modperl
>scripts). They both output a large HTML file.
>Server #1 run and outputs the data fine but produces many warning on
>the error log.
>Server #2 freezes somewhere on the way. Since output warnings to the
>error log are buffered (I know this from just watching the error log
>with tail). I guess that the whole scripts freezes because the buffer
>gets full, right?
>So, first of all, can someone confirm this?

If your theory is correct, it's likely that the server process will go
into a "blocked" state when it makes the write() call to push the data
into the buffer. If you watch the CPU usage of this process in top or
a similar tool and see that it is high (nearing 100%), it's more
likely that your software has got itself into an infinite loop
somehow.

When a process is blocked it isn't given any runtime by the kernel
scheduler, so its CPU usage sould be zero.

Good luck,
-Claire