Hi,

1st post here. I had to replace my parity drive. I ran the pre-clear script twice before adding it to the array. I ran the "New config" script and started the array with the new parity drive. It ran the parity check and found 206 errors. When I check the log I see a lot a errors on one of my data disk:

 

Sep 22 18:29:43 Tower kernel: md: disk4 read error (Errors)

Sep 22 18:29:43 Tower kernel: handle_stripe read error: 3469467840/4, count: 1 (Errors)

 

and

 

Sep 22 18:29:37 Tower kernel:          res 51/40:ef:00:e1:cb/00:02:ce:00:00/f0 Emask 0x9 (media error) (Errors)

Sep 22 18:29:37 Tower kernel: ata3.00: status: { DRDY ERR } (Drive related)

Sep 22 18:29:37 Tower kernel: ata3.00: error: { UNC } (Errors)

 

What should I do next? Run another parity check? Run reiserfsck? Replace the drive?

 

I attached the smart report of the drive giving me errors. I had to zip the files, my syslog was to large to attach.

syslog_and_smart.zip

Your disk has three  un-readable sectors.  If you perform another parity check now, they should be re-constructed from parity.

 

Unfortunately, since you just calculated parity with contents of those sectors (presumably zeros returned from failure) , those three sectors will probably be zeros when re-constructed.

197 Current_Pending_Sector  0x0032  200  200  000    Old_age  Always      -      3

 

About all you can do is perform another parity check (first) followed by a reiserfsck on /dev/md4  Unless those three sectors were part of the file-system itself the file system check should be fine.  (The sectors are far more likely to be unused space on the disk or part of some file stored on the drive)

Edit: see next post...

 

Keep an eye on the disk.  Most disks have several thousand spare sectors to be used for re-allocation, and if the number does not grow dramatically over time, the disk should be fine.

 

 

 

Joe L.

Your disk has three  un-readable sectors.  If you perform another parity check now, they should be re-constructed from parity.

 

Unfortunately, since you just calculated parity with contents of those sectors (presumably zeros returned from failure) , those three sectors will probably be zeros when re-constructed.

197 Current_Pending_Sector  0x0032  200  200  000    Old_age  Always      -      3

 

About all you can do is perform another parity check (first) followed by a reiserfsck on /dev/md4  Unless those three sectors were part of the file-system itself the file system check should be fine.  (The sectors are far more likely to be unused space on the disk or part of some file stored on the drive)

 

Keep an eye on the disk.  Most disks have several thousand spare sectors to be used for re-allocation, and if the number does not grow dramatically over time, the disk should be fine.

 

Joe L.

Just noticed....  looked more closely in your syslog.

Sep 22 10:56:56 Tower kernel: mdcmd (22): check NOCORRECT

 

You did a non-correcting parity check...  In that case, parity probably has the correct information for the three un-readable sectors.

 

You can stop the array, un-assign disk 4, start the array with it un-assigned, then stop the array once more, re-assign it, and start the array letting it re-construct the drive.  It will then be able to write the correct contents of those sectors to the drive while re-constructing it.

 

Joe L.

Thanks Joe for all infos. I'll try to rebuild my drive tomorrow and I'll report back the results.

 

Chuck

 

Sent from my Nexus 7 using Tapatalk 2

I finally had the time yesterday to work on my server. I'm almost done rebuilding the drive. So far it's looking good.

 

My question now is how to make sure that my parity is good?

 

 

 

 

 

Sent from my Nexus 7 using Tapatalk 2

Run parity check.