I need help with a strange problem.
We currently have c3750 12 & 24 port ipbase image closet switches that connect to
a stack of 4 c3750-12s ip enhanced imaged switches. The closet switches connect to
the distribution switches via cross stack layer 2 etherchannel the 3750-12s distribution
stack connects to a 6506 core switch via cross stack layer 3 port channel.
Here is my problem :
Host A & B are on the same closet switch, same subnet and same vlan. Host A can ping host C
which is on the other side of the 6506 switch and Host B cannot. Both A & B can ping
another host on C's subnet. When running a trace route on problem the packets stop at the
12s distribution switch. Yet the 12s distribution switch can ping both Host B and C.
There are no ACL's and utilization is minimal. The proper routes are in the routing table
and there are CEF entries for the destination.
Cisco TAC is Looking at the issue but is stumped so far. The problem actually includes multiple
source and destination addresses and any connectivity to them not just pinging but the example
is the problem in a nutshell. We have to shutdown one of the layer 3 port channel link so
the problem does not return. (it doesn't matter which link).
Clearing routing table, turning CEF off, changing load-balancing algorithm on switches to
ip src-dst, changing master switch in stack to switch #2, rebooting stack (which will clear
the issue for a day to week or so ). Cisco has checked the config of thes switches and they
anyone seen anything like this?
Our next option is to upgrade IOS on the 12s's .