[users@httpd] Solved was Re: [users@httpd] Apache taking (exactly) 30 seconds to serve static images

[oh...@cox.net]

---- "Macks wrote: 
> Can you run either strace (or dtrace depending on platform) against the running process and see what it is doing during the request?
> 
> A
> --
> Aaron Macks
> Sr. Unix Systems Engineer
> 
> Harvard Business Publishing
> 300 North Beacon St.    |   Watertown, MA 02472
> (617) 783-7461                |   Fax: (617) 783-7467
> www.harvardbusiness.org |   Cell:(978) 317-3614
> 
> On Jul 16, 2012, at 6:30 PM, <oh...@cox.net>
>  wrote:
> 
> > 
> > ---- Vivek Nambiar <vi...@gmail.com> wrote: 
> >> May be you can add the hostname and ipaddress in the linux hosts
> >> file,restart apache and see if that makes any difference??
> >> 
> >> Thanks
> >> VIVEK
> >> 
> > 
> > Hi Vivek,
> > 
> > I made suer that there was an entry in /etc/hosts, with both the FQDN and short hostname and bounced Apache, and still have the same problem.
> > 
> > Again, I'm wondering what might be the difference re. why does it work fine when I run Apache in single process mode, but then it has this stalling problem when I run it normally?
> > 
> > Jim
> > 
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> > For additional commands, e-mail: users-help@httpd.apache.org
> > 
> 

Hi,

For the record, I think that we were able to eliminate these problems.  There were a number of things:

- We had to add in the httpd-ssl.conf:

KeepAlive Off

- Also, I noted that when I ran "mount" it was showing that the SSLSessionCache and the semaphone/mutex directives in the httpd-ssl.conf were pointing to a mount point on which a mapped volume was mounted, rather than a physical drive.  Also, the mount point didn't have "nosuid".  So, I changed both directives to point to /tmp

After we bounced the Apache, it worked, both in single-process mode and in normal mode.

The original hint was that it would run in single-process mode, so I guessed that the problem when in non-single process mode was related to some mechanism that the processes were using to coordinate, which lead me to the SSLSessionCache and semaphore/mutex directives.

Thanks,
Jim

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org