HBA Failure

PowerPath could be helpful in detecting flaky hba/cable/gbic as it will start logging errors that could be seen from "powermt display" output. You can also run some solaris commands to see if it's reporting any failures:

http://docs.sun.com/source/819-3741-13/8_V445_Diag.html#50630688_62115

In general if there is a bad cable/hba/port ../var/log/adm/messages will start filling up with disk/scsi errors and that should be a good indication to your system admin that's something is not right.

dynamox,

can you please specify the outputs or possible outputs in scenarios like so that we can figure out what the issue at that point itself.

1. Bad HBA,GBIC, Cable
2. HBA is fine, problem with the switch or zoining, switch hba is down.
3. Symm director port down, director offline, SP port down.

unfortunately i don't have anything on hand to share with you ..fortunately these problems do not happen very often but if they do the host will start spewing scisi/powerpath errors to the log. If you have a test system available, dual attach it, present a LUN, generate some IO to it. ..and then pull one cable out. That should generate tons of messages in syslog.

On the LINUX hosts i check "Commands retried with dropped frame(s) = xxxx" from "cat /proc/scsi/qlaxxxx/y". it is always good to start by swithcing GBIC module/port , then cable if both of them did not solve the problem then replace the HBA.

SKT,

Can you kindly explain little more, /proc/scsi/qlaxxxx/y what are the xxx and y, by switching GBIC module/port did you mean by re-seat the HBA.