IBM SAN768B-2, Руководство пользователя

Страница: 198 / 228

3649 or FC3633) then you should not continue the tuning guidelines outlined in
this section. Try other troubleshooting options or call IBM Support for additional
help.
If the error found is a C2-1010 message or a C3-1010 message originating from a
core blade, then check to see whether the bit errors that are occurring are those for
which tuning can help with. This can be done by looking for C2-5825 or C3-5825
event messages that triggers when a CRC error with good EOF (crc_good_eof) is
detected on an internal port.
Note: Cx-5825 events can only be viewed by logging in as the root account to view
additional error logging.
The Cx-5825 event message has the following format:
Message <timestamp>, [Cx-5825], <sequence-number>,, WARNING, <system-name>,
S<slot number>,B<blade port number>: Detect CRC error with good EOF on C-port:
<blade port number> count <error count>
An example of finding a C2-5825 message using the
errdumpall
command is
shown below:
errdumpall | grep C2-5825
2013/03/05-01:01:01:307873, [C2-5825], 2/0, SLOT 6 | CHASSIS, WARNING,
testsystem, S3,B-1: Detect CRC error with good EOF on C-port(15)
count(
21300 ), OID:0x4332800f, ...
2013/03/05-02:02:01:307873, [C2-5825], 2/0, SLOT 6 | CHASSIS, WARNING,
testsystem, S3,B-1: Detect CRC error with good EOF on C-port(15)
count(
21324 ), OID:0x4332800f, ...
Identify the slot that is having an issue in the error message by noting the S<slot
number>, B<blade port number>. In the above example, the S<slot number> is S3,
meaning that the port blade in slot 3 potentially has the issue, and blade port
number is 1. The error message includes a timestamp with the format
<year>/<month>/<date>-<hh>:<mm>:<ss>:<ms>. The error message also includes
a count(#) field where # is a count of the number of errors logged since the counter
started.
The error message for Cx-5825 should be polled over a period of time while traffic
is running to see if it is incrementing for a particular slot and blade port number.
This should ideally be done over the interval when the application is experiencing
performance problems in order to see if the issue is related to internal CRC errors
with good EOF.
1. If the Cx-5825 counters are not incrementing for any specific slot and blade
port number then this indicates the Cx-1010 error message was logged for other
types of errors for which tuning may not help. Contact IBM Support for further
assistance.
2. If the Cx-5825 counters are incrementing for a specific slot and blade port
number then tuning for the internal links may help resolve the issue. Make a
note of which slot number is experiencing issues.
In the C2-5825 error message examples above, the timestamps for the first error
message is 2013/03/05-01:01:01:307873 and 2013/03/05-02:02:01:307873 for the
second error message. Approximately 60 minutes elapsed between when the first
and second messages were logged. The count of the first error message is 21300
and 21324 for the second error message. This indicates approximately 24 CRC
errors with good EOF were detected over this 60 minute period.
170
SAN768B-2 Installation, Service, and User Guide