Users Guide

Fabric OS Troubleshooting and Diagnostics Guide 89
53-1002930-01
Using the spinFab and portTest commands
10
• Verify that the cables and SFPs are inserted properly. Remove and insert them again on both
ends.
• Verify that the failing local port is working when connected to another remote port. Similarly,
check whether the failing remote port is working when connected to another local port.
• Once the fault is isolated on either the local port or the remote port, replace the cable and
SFPs connected to the local port and the remote ports.
• In case of loopback ports, change the loopback plug. Refer to “Marginal links” on page 30 for
more information on changing the loopback plug.
• Further isolation can be done by running the portLoopbackTest command (Offline test) on the
failing port to check whether the blade internal ports are having some problems.
- The -lb_mode 1 operand verifies that the SFP is working within normal operating
parameters. The use of this operand requires that loopback cables are connected.
- The -lb_mode 2 verifies that the ASIC port is working within normal operating parameters.
The use of this operand does not require any loopback cables.
Tx/Rx errors
The following errors are seen when the port fails to transmit or receive the frames.
ERROR: DIAG PORTSTOPPED spinfab:spinfab, 0 nMegs,
Pt7/2(2) Ch0/2 No Longer Transmitting, FTX Counter Stuck at 116295726,
ERROR: DIAG TIMEOUT spinfab:spinfab, pass 2,
Pt0/17(7) Ch0/7 Receive Error/Timeout
If you receive any of the Tx/Rx errors, follow the suggest debugging procedures:
• Check whether the same port is reporting link errors as discussed in “Link errors” on page 88.
If yes, follow the same set of debugging procedures as discussed in “Link errors” on page 88.
• Check whether the local port or the remote port is beyond port 255. If yes, try connecting to
the lower number of ports. This behavior is found in Fabric OS v6.2.0 and earlier versions only.
• Check whether the local port or the remote port port is part of a shared-area region. If yes, try
connecting to the non-shared area ports. This behavior is found in Fabric OS v6.2.0 and earlier
versions only.
• Check whether the local port or the remote port is having its area swapped. If yes, try
connecting to the normal area ports. This behavior is found in Fabric OS v6.2.0 and earlier
versions only.
Clearing the error counters
This procedure clears the port hardware statistics, including ALPA-based CRC monitor, End-to-End
monitor, and filter-based performance monitor statistics.
1. Connect to the switch and log in as admin.
2. Enter the portStatsClear slot/port command.