[ic] Payment Module 500 Error

[Ron Phipps]

> From: Chris Ochs

Chris, Please don't top post... it makes it difficult to follow a thread
and a lot of people won't respond if they see lazy quoting.

> This sounds a lot like the problem I am having with our own payment
> module.
> If you modify the boa module to log the actual response LWP is
returning,
> I'll bet it will say 500 read timeout.  On my system it returns this
in 1-
> 3
> seconds.   What I think it happening is that the LWP is aborting
> prematurely
> upon catching some signal that it shouldnt' be.

This sounds like a problem with Perl signals.  Here is a previous view
of a very similar/same problem.

http://www.icdevgroup.org/pipermail/interchange-users/2002-March/018636.
html

> Also, in rpc mode this doesn't happen to me, just in low or high
traffic
> modes.  You might try running in rpc mode just to see if the problem
goes
> away.  If it does, then at least we would have more evidence that the
> problem isn't just something on our individual systems, but will
probably
> effect a lot of people.  I think more people haven't reported this
because
> a
> lot of production systems probably haven't upgraded recently.

The reason why most this doesn't happen in RPC mode is because of the
"MaxServers 0" setting which is now the default for that traffic mode.
If you were to add MaxServers 0 to low or high mode it would probably
work as well, EXCEPT that this could/will have side effects, so please
leave it in RPC mode.  If you were to remove MaxServers 0 from the rpc
mode section the problem would show up again.  I'm pretty sure not all
the catalogs out there are running RPC mode, so it does not affect
everyone.  But the fix simply is to switch to RPC/MaxServers 0 for
4.8.x-5.0.x

> I am not sure this is an interchange bug per say.  An older install of
> interchange I have on the same OS and perl version works fine, the
> difference is that several of the perl modules are a newer version.
This
> might very well be a bug in LWP or it's dependencies that cropped up
just
> recently.
> 
> Chris

It's a combination of OS/Perl/IC.  Taking a catalog that was running
fine on 4.6.x and upgrading it to 4.8.x on the same system with the same
modules showed this problem.  Although we never ended up testing the
RPC/MaxServers 0 setting on that particular catalog my guess is it would
fix the problem since it has done so on so many others.  Mike's words
were that this was a well known Perl signaling problem that happens to
show up when IC calls external programs/modules.  We saw it happen with
calls to Verisign (which doesn't use http/https to post), Sendmail, pgp,
etc.  It also happens more often on SMP systems.

Good luck!
-Ron

> 
> ----- Original Message -----
> From: "Matt Johnston" <matt at thebackpacker.com>
> To: <interchange-users at icdevgroup.org>
> Sent: Saturday, January 17, 2004 9:39 AM
> Subject: [ic] Payment Module 500 Error
> 
> 
> > I am using the BoA payment module for Interchange 4.9.9 and I am
> constantly
> > getting errors where Interchange thinks that the payment module has
> erred
> > but BoA has completed the transaction.  A good percentage of the
time
> > (close to 50%) when I post a transaction to BoA I get a "500
(Internal
> > Server Error) read failed" error.  I have contacted BoA about this
error
> > and they say that it is on my side.  I am not very well versed in
perl,
> so
> > I am having a hard time tracking this down.  The BoA support said to
> check
> > for a timeout setting (BoA uses 120 seconds) and to make sure that
the
> > KeepAlive header is not present.  I have searched the code to the
best
> of
> > my ability but cannot find any references to these in Interchange.
> >
> > This led me to believe it may be some version of a program on my
server
> > that is causing some problems.  Here is my setup:
> >
> > RedHat 8
> > perl 5.8 installed using sources and the instructions on icdevgroup
> > openssl 0.9.6b-33 from RedHat rpm
> > Crypt:SSLeay 0.51 from sources
> > LWP:UserAgent 2.24 from sources
> > apache 2.0.47 from sources
> > Dual Pentium Processor
> >
> > Can anyone think of something that may be causing this problem?
> >
> > Matt Johnston
> >
> > http://www.lattaoutdoors.com
> > http://www.thebackpacker.com
> >
> >
> > _______________________________________________
> > interchange-users mailing list
> > interchange-users at icdevgroup.org
> > http://www.icdevgroup.org/mailman/listinfo/interchange-users
> >
> 
> _______________________________________________
> interchange-users mailing list
> interchange-users at icdevgroup.org
> http://www.icdevgroup.org/mailman/listinfo/interchange-users