[c-nsp] 6509 problems

[Paul A]

Hi all.

We have been using a 7200 and recently we purchased a 6509. The plan is to
move everything to the 6509 from the 7200. In order to accomplish this our
plan was to connect the 7200 to the 6509 via GigE.

Our current network is like this:

Internet traffic hits the 7200, the 7200 has a riverstone switch directly
connected to it via fastethernet which handles a bunch of vlans and then we
have the 7200 connected to the 6509 via GigE.

On the 6509 we moved out www clusters to it, which are connected to another
GigE interface on that switch.  The www cluster was previously connected to
the 7200 that same way its connected to the 6509 without any issues. The
cluster is connected to the 6509 on another GigE interface. This interface
has multiple ip, secondary ips and is a routed interface which connects to
an HP switch that the www clusters are on.

We have hard coded speed and duplex on all interfaces and we have 0% packet
loss etc. What is happening is that customers from various points on the
internet cannot bring up sites on the cluster on port 80 and probably other
ports.  These customers cant even telnet to port 80. They can however ping
and trace everytime when the port 80 issue happens ( I have no ACLS). With
that said other customers from other point on the internet can pull the same
sites without any issue, again I have no ACLS that would do this. 

Some of the symptoms customers report, not everything comes up on the site,
can't telnet to port 80/25/etc on the server but can ping/trace to it.  From
my desktop that is directly connected to a switchport on the 6509 I never
have that problem.

I have look at cef/arp/acls/etc and everything looks fine from a config
prospective but I still can't figure out the problem.  Although when I do a
shut/no shut on the GigE interface on the 6509 that is facing the cluster
customer report that after that they can pull up the website again.

Thanks in advance, Paul