The Role Of Memory Scanning And Memory Mapping - Extreme Networks BlackDiamond 6804 Troubleshooting Manual

Diagnostics
• Background transceiver scanning
— Tied to the system health check configuration
— Runs in background to detect potential control path faults
— Tests internal transceiver data paths
— Tests all ASICs for proper read/write operations

The Role of Memory Scanning and Memory Mapping

The memory scanning and memory mapping functions identify and attempt to correct switch fabric
checksum errors. When you are in the process of implementing the ExtremeWare diagnostics, keep in
mind that these functions are an underlying base for much of what takes place in the diagnostic tests
that make up the system health checks diagnostic suite. For more information, see Chapter 3, "Packet
Errors and Packet Error Detection."
NOTE
Memory scanning addresses switch fabric checksum errors detected in the packet memory area of the
switching fabric.
The ExtremeWare memory scanning and memory mapping diagnostics are analogous to hard disk
scanning tools, which are used to detect and map out bad sectors so that the drive can remain
operational with no adverse effects on performance, capacity, or reliability. The ExtremeWare memory
scanning and memory mapping diagnostics are used to identify and correct switch fabric checksum
errors.
Memory scanning and memory mapping are two separate functions: scanning detects the faulted
portion of the memory; mapping re-maps the memory to remove the faulted memory section.
Memory scanning is designed to help isolate one of the major root causes of fabric checksum errors:
single-bit permanent (hard) failures. Memory scanning detects—with a high probability—all current
single-bit permanent (hard) failures in the switch memory that would result in fabric checksum errors.
Memory mapping can correct up to eight of these detected permanent (hard) single-bit errors by
reconfiguring the memory maps around the problem areas.
The packet memory scan examines every node of packet memory to detect packet errors by writing data
to packet memory, then reading and comparing results. The test is invasive and takes the switch fabric
offline to perform the test.
Errored cell correction:
• If the test detects eight or fewer error cells, those error cells will be mapped and excluded from use.
The module will continue to operate.
• If the test detects more than eight error cells, the module is identified as "failed" and must be
replaced.
You should use this test when the system log displays some intermittent or sporadic error messages that
might indicate a problem, but do not provide sufficient information to confirm the problem or isolate
the fault.
52 Advanced System Diagnostics and Troubleshooting Guide