Brocade Network OS Administrator’s Guide v4.1.1 - Manual (Page 682)

Verifying SEEPROM data
1. To verify the SEEPROM, copy the test_symod file to /fabos/bin as test_sysmod , and select option
10 for i2c and option 27 to Verify FRU Seeprom. The test begins automatically.
2. Use the offset of 0x6a4c, as that is where the IP table starts (size 256), but any offset (and size less
than or equal to 256) will access that device.
CPU use is unexpectedly high
Unexpectedly high CPU use is usually the result of a process consuming a large percentage of
available CPU cycles. It can prevent access to the switch by Telnet or make an ISL nonfunctional.
If you suspect high CPU use, complete the following steps.
1. In privileged EXEC mode, enter the show process cpu command to determine which process is
causing the high CPU reading.
2. Shut down the corresponding interface or delete the configuration suspected of causing the high
CPU use.
ECMP not load balancing as expected
Equal cost multipath (ECMP) routing increases throughput by balancing traffic across multiple routes
that tie for best cost. If you suspect that traffic is not being balanced as expected, complete the
following steps.
1. In privileged EXEC mode, enter the show fabric route topology command to display whether
ECMP routes are expected.
switch# show fabric route topology
Total Path Count: 1
Src Dst Out Out Nbr Nbr
RB-ID RB-ID Index Interface Hops Cost Index Interface BW Trunk
-----------------------------------------------------------------------
66 1 124 Fi 66/0/4 1 500 129 Fi 1/-1/-1 32G Yes
If the output shows multiple equal-cost paths between the source and destination switches, then
ECMP load balancing is expected.
2. Check the interface utilization to verify whether it matches with the expected number of flows.
3. Enter the l2traceroute command to investigate whether Layer 2, Layer 3, and Layer 4 flows hash to
separate ECMP links.
To avoid disruption of operation inherent in ECMP, the correctly functioning Brocade routing
strategy routes a specific flow along one deterministic route. Additional flows take available equal-
cost routes. This step verifies whether this flow hashing strategy is functioning correctly.
For details about using the l2traceroute command, refer to Using Layer 2 traceroute on page
714.
ENS not working correctly
The Ethernet Name Server (ENS) is working correctly when the content of MAC address tables is the
same among switches in the same VCS Fabric cluster. Perform the following checks to ensure that
ENS is working correctly:
Verifying SEEPROM data
682 Network OS Administrator’s Guide
53-1003225-04