uncorrectable read errors

[Scott_20]

If i get this in syslog for a particular disk, write as well, is there a command i can run from the command line to test the disk? thanks

[A. Clay Stephenson · Accepted solution]

You could do something like this as a safe read test:

dd if=/dev/rdsk/c1t5d0 bs=256k of=/dev/null

although you would be better served by using
STM.

If you are seeing those kinds of errors in syslog, you would be well-advised to believe them. By all means check termination, terminator power, bus length, ... --- but when that has been eliminated, it's really time to replace the disk before it fails completely. If you are mirrored, this is a complete non-event; if the disks are hot-plug and mirrored then you don't even need to shutdown.

[SS_6]

Do this with no impact to disk
dd if=/dev/rdsk/c1t5d0 of=/dev/null bs=512k
You can also do pvdisplay on that disk and see all PV's are in current status.
#pvdisplay -v /dev/dsk/cxtxdx

[generic_1]

I think stm would be your best helper in this situation. Also, take a look at your /var/adm/syslog/syslog.log for the disk error. If your monitoring tools made an entry you should be able to cut and paste that string with a -a at the end and it should not only give you output about the error, but possible causes that you may want to investigate.

dmesg can also be sligtly helpful in seeing errors, but the history is short on it and it is not time/date stamped.

The dd commands mentioned above will basically try to read data from all of parts of the disk and if there is any problems will stick out due to read errors.

If your disks are external it would not hurt to check your connections and make sure no one was doing any changes that could have impacted your disk temporarily.


Good Luck

[Bharat Katkar]

hi,
dd if=/dev/rdsk/c1t5d0 bs=256k of=/dev/null

this will serve your purpose.
Regards,