Parity problems are common as lightning strikes and like a lightning
strike, chances of it striking the same location is slim. Parity
problems have plagued systems that use this type of memory for decades
now and there is no real fix. The unfortunate thing about this parity
error bit flip is a problem that affect any computing system (that uses
the parity mem). As to what they happen is an unsolved mystery for 40+
years (since early 70s).
Essentially, there are 2 types of parity errors. "Soft" parity errors
occur randomly and rarely and are often the result of a transient issue.
These may show up one time and never be seen again. These usually
require the box to be monitored for further incidents. "Hard" parity
errors may occur frequently over a short period of time. When these
errors are seen repeatedly, this often indicates a hardware problem and
would require an RMA. Soft parity errors have been proven to be 10 to
100 times more frequent than hard errors.
Another point on software parity errors is that Cisco devices are capable of correcting some errors, I have seen switches reporting that it detected & corrected a parity error, i have seen this lot of times on 6500. What Cisco really recommends is that if the switch keeps on reporting this parity error for multiple instances, get the device/module or memory part replaced. This will be decided by looking at the complete "show tech" and seeing if there is a crash/traceback associated with the hard parity error.
HTH,
Amit Aneja
