Hard error found on the hard disk

[mikecalizo]

Hi All,

I am about to do solaris patching. During pre patch checking I found error on the hard disk. Is it still safe to proceed since I am going to break the mirror and create a new BE using c1t1d0.

#iostat -En
c1t1d0 Soft Errors: 0 Hard Errors: 3 Transport Errors: 0
Vendor: SEAGATE Product: ST373405FSUN72G Revision: 0638 Serial No: 0227K1WG96
Size: 73.40GB <73400057856 bytes>
===================================
d12: Submirror of d10
State: Okay
Size: 16780224 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t1d0s0 0 No Okay

d32: Submirror of d30
State: Okay
Size: 16780224 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t1d0s3 0 No Okay


Thanks in advance.

 

[djr111]

has the drive been replace prior, if so maybe they are old errors from past disks. i believe these errors in iostat stay till the server is rebooted.

If the disk has not been replace. I would not trust it, but that is me just erroring on the side of caution. I really don't like making more work for myself if I do not have to.


what does format or /var/adm/messages say? it looks as if metastat does not see any errors, but it does not always get them right away.

 

[mikecalizo]

Hi djr111,

I can not find any logs related to the disk error from /var/log/meassages as well as on dmessages because there is an application running and it keeps updating the logs.

The server was up 56 days and I am sure no disk repalcement was done within that 56 days. Metastat is not saying any "Needs Maintenance" error for both slices (c1t1d0s3 and c1t1d0s0).

Should I reboot the server first and see if error are still there or should I replace the disk right away and then reboot?

Is there a way to verify this error without reboot?

Tnx.

 

[dandan123]

You can always do a format - read check and see if it shows any errors.

 

[djr111]

I guess it depends on how careful you want to be. The most careful way would be to replace and continue on. however, you could try to reboot. but that will only just clear the iostat stats, if they were actually true, they would be gone, but your problem may now be masked. If you intend on doing any maintenance, my only advise would be to break the mirrors and do what you need to do.

 

[KenCunningham]

I concur with djr111, but take a backup in any case!

I want to be good, is that not enough?

 

[mikecalizo]

Hi, I proceed by doing the patching. The server is now up and running. What I did is I break the mirror for d10 and d30 which makes d12 and d32 an untouched slice and serves as backup in case. Reboot the machine to Single user mode and do patching. Currently we are observing the application.

Thanks for the help.