[ic] mod_interchange and Apache MaxClients

[Ron Phipps]

> From: interchange-users-bounces at icdevgroup.org
[mailto:interchange-users-
> bounces at icdevgroup.org] On Behalf Of jeff at hisgirlfridays.com
> Sent: Thursday, November 17, 2005 5:45 AM
> 
> On Thu, Nov 17, 2005 at 10:46:37AM -0000, John1 wrote:
> 
> > In effect, I think I have verified that Apache doesn't start
responding
> > again even if port 80 were to be blocked.  Once MaxClients has been
hit
> > Apache stops servicing requests as it won't spawn any more processes
and
> > for whatever reason non of the Apache processes ever die after this
> point,
> > even if the server is left for many hours over night when I know the
> > traffic to my site is very low indeed.  Ron & Jeff - do you also see
> this
> > effect?
> 
> Yes, none of the servers die.  I can also add that the event is
> predicated by mod_interchange and IC failing to communicate and
holding
> apache children open.  During this initial period, the web server will
> serve non-IC content but any requests for IC pages "freeze up".  As
> requests for IC pages stack up, MaxClients is eventually reached and
> apache stops serving requests entirely.
> 
> One of the reasons I never mentioned the issue before is because I am
> running the IC 4.8 series .  I am pretty sure a lot of the
> process-forking code was improved between 4.8 and the current version
> and those improvements could very well resolve the issue.

I'm running the 4.8 series as well and unfortunately we have a lot of
custom code in the server portion that we'd need to break out if we were
to upgrade IC at this point.  We are doing a complete site rewrite that
is based off of 5.3.x, so we are trying not to do two upgrades if
possible.

We also cannot run in High traffic mode to see if the traffic mode is
part of the problem because in low and high traffic mode we get double
charges and sendmail/pgp can fail.

Jeff and John have you tried high or low traffic mode to see if the
issue goes away in those modes?

> > I don't think that the issue is anything to do with my Virtual
Server
> > environment, but I thought I better throw it into the mix.
> 
> On my side, the issue has spanned multiple physical servers with
> completely different hardware and multiple versions of linux and
debian.
> 
> 
> Jeff

We've run IC on about 10 systems over the years.  The server I question
has been up for about 3.5 years and has been running mod_interchange for
the last year or so and has not experienced this problem up until 1 week
ago.  It's strange that a server would run solid all this time then all
of a sudden start exhibiting this behavior.

-Ron