[mythtv-users] New HD recommendation

[faginbagin]

On 4/18/2016 6:55 PM, Douglas Sargent wrote:
> Hi all,
>
>          I have 4 x 2TB disks which have been up and running the last 3
> years nonstop, with no issues... cross your fingers.
>
>         Can someone recommend big disks that are reliable.
>
>         Once I buy them. Will it be as simple as mount it, copy
> everything across, unmount the old, the mount the new with the name of
> the old one. Seems right but way to easy. Darn I like Linux.
>
>        thanks guys
>
> Doug

I look for drives with the longest warranty. This Christmas I bought 4 
3TB HGST drives with 3 year warranties. Got to look close, because some 
only have 2 year warranties. Before I put any new drive into service, I 
use the following procedure to increase the chances it's a good drive 
that will last.

Capture SMART data via:
smartctl -a /dev/sdx > smart0.out

Write semi random data on the whole drive with:
nohup shred -n1 -v /dev/sdx > shred.out &
Go away for a couple of hours. Check shred.out and figure out how long 
it will take to finish. Come back when it should be done.

Read the whole drive and compute a checksum:
nohup md5sum /dev/sdx > md5sum.out &
Go away for roughly the same time it took shred to write to the drive.

Read the whole drive again and make sure the checksum matches:
nohup md5sum -c md5sum.out > md5sum.chk &
Go away for roughly the same time it took the first md5sum to read the 
drive.

Write zeros to the drive:
nohup dd if=/dev/zero of=/dev/sdx bs=1M > dd.out &

Capture SMART data via:
smartctl -a /dev/sdx > smart1.out

Compare the two smart runs:
diff smart0.out smart1.out
Make sure there are no complaints about sectors.

Make sure the kernel didn't report any errors:
dmesg| tail

If no SMART or kernel reported errors, create partition table, create 
partitions, mount, etc...

If any errors, return immediately.

Regards,
Helen