[ic] Verisign PFPRO Transactions Getting Dropped

[Ron Phipps]

> From: interchange-users-admin@interchange.redhat.com
[mailto:interchange-
> 
> > (context lost due to lazy top-posting followups throughout the
thread)
> >
> > we have posted this similar problem before as seen in some previouse
> threads there has not been a soultion that we have
> > come up with.  we have tried numerous things like the prefork mode
that
> was mentioned and that did not help us.  Our
> > server doesnt have that heavy of a load on it.  This is the only
site
> running on it and it does not get that much
> > traffic.  I really doubt if it is an IO issue on our situation, but
I
> will look into that a little more.  It did work
> > when I was using 4.6 it does have this problem when I switched to
4.8.3
> >
> Has anyone considered increasing the timeout value from 10 seconds to
> a higher value.  If Verisign can't afford machines that are quick
> enough to respond in a reasonable time, then perhaps an increased
> timeout, or another provider, will help.
> 

Kevin I do not think this issue has to do with how quick verisign's
server are.  We tried a timeout of 30 seconds and the problem still
occurred.  The problem did not exist in 4.6.1 and only showed up when
the catalog was brought up on 4.8.x (not implying there is a problem
with IC 4.8.x).  At the same time we also received problems with
Sendmail and PGP failing under load (one other person has seen this
too).  Sendmail reported: /usr/sbin/sendmail: could not send mail. And
PGP reported: PGP failed with status: (the out of memory code).  All
errors did not show up when not under load. And the errors were
intermittent.

Reverting back to 4.6.1 and the problems were no longer seen (on the
same hardware).  Can anyone think of something that may have changed
between 4.6.x and 4.8.x in the way external programs are called?  The
only common thing I see between people reporting this problem (I think
only 3 of us have) is that the catalog worked in 4.6.1 but no longer
works in 4.8.x.  Maybe it's something in our catalog.cfg that chokes on
the new server?

The problem has now been seen in the following environments:

Redhat 6.2, 7.1
IC 4.8.1, 4.8.3
Verisign 2.0, 3.0
Verisign/IC Module Old signio and new Signio.pm
Perl 5.003, 5.6.0
Single and Multi processor

In our case the server should have had enough power to serve the
catalog.  It's a Dual 500, 512mb ram, 9GB SCSI drive serving only one
catalog with 400 products under light load.

We were forced to revert back to IC 4.6.1 but we will be tackling this
problem again in the near future.  Any ideas for tracing what is going
on when this error occurs?  We turned on debugging and logged the Signio
and PGP calls, they all looked as expected.

The mystery continues ;)
-Ron