Troubleshooting guide

E-8
ATM and Layer 3 Switch Router Troubleshooting Guide
OL-1969-01
Appendix
Architecture Problems
Note The unavoidable side effect of this type of failure is that the console appears to hang. You might think
the console is not responding, and take drastic actions to recover. Unfortunately, your actions might
leave you with little data to determine the root cause of the problem. In other words, what might
appear to be a console failure could be merely a significant delay on the part of the console to reply
to some commands.
For example, after the initial error message appears, the console screen might completely fill with error
messages similar to those in the following example:
%AAL5-3-INTERNAL_ERROR: aal5send: mmc_queuecell failed queue cell fail output port
FastEthernet0/0/3 input vpi 0 input vci61
queue cell fail output port GigabitEthernet2/0/1 input vpi 0 input vci154
queue cell fail output port GigabitEthernet2/0/1 input vpi 0 input vci155
queue cell fail output port FastEthernet3/0/0 input vpi 0 input vci172
queue cell fail output port FastEthernet3/0/1 input vpi 0 input vci174
queue cell fail output port FastEthernet3/0/2 input vpi 0 input vci176
queue cell fail output port FastEthernet3/0/3 input vpi 0 input vci178
queue cell fail output port FastEthernet3/0/4 input vpi 0 input vci180
.
[Information Deleted]
Potential Causes
An mmc_queue cell or a port stuck instance could be either a hardware or software issue. The hardware
problem could be a single port failure, or it could be a failure in one or both of the switch-fabric or
switch-processor modules.
The failure might occur only on a single port. However, if this port provides significant data flow and
many other ports send traffic to it, the bad port can impact all other ports that communicate with it (for
example, a port connected to a heavily utilized server or a port connection to the backbone). Since
port-to-port communication within the Layer 3 enabled ATM switch router uses static PVCs, a single
port failure can cause cells from all other ports to be backed up on the incoming VC to this port.
The cells queued up on one or more VCs used for port-to-port communication can be shown by entering
the show epc queuing command.
The following are some commands that can be used to determine whether the port stuck instance is a
hardware failure on one port, a module failure, or a failure in one or both switch processor modules:
• Switch processor module failure:
– The show epc queuing command lists many ports from different interface modules having 300+
cells queued up, and some have more than 500-600+ cells queued up. The X-INT might point
primarily to a switch route processor listing only a few cross-connections to other physical
ports.
– The show mmc ports command helps determine whether the ports in the output of show epc
queuing command are mapped to either even MSCs (0,2,4...) or odd MSCs (1,3,5...). If all ports
map to an even MSC port, then the problem might be a failure of the even switch processor
module in slot 5, which contains the even MSCs. If all ports map to an odd MSC port, then the
problem might be a failure of the odd switch processor or slot 7, which contains the odd MSCs.
If the ports map to both MSCs, but the commonality still consists of many VCs from different
ports going to the switch route processor or the route processor module, then the problem could
be both the switch processor modules.
– The show epc status command might display “not OK” in the status field for most of the ports.