When drives get to their end of life, the heads are worn out and don't 
"see" very well the difference between a 1 and a 0. If this is detected 
[by CRC f.e.] the drive winds back and does the read or write again. It 
will give up after several tries, thereby reporting a fatal error. It 
could be that this re-checking is happening.

On some OS's you can run iostat -E and observe "soft" or "recoverable" 
errors. If this is non-zero, that's what's happening.

HTH, Joost

Richard Hellier wrote:
> We have an ATL P3000 library with 5 DLT drives.  Just recently, backups that
> happen to land on two particular drives among the five seem to run much more
> slowly than expected. Anyone else seen anything like this, please, and can
> suggest a remedy?
> 
>  
> 
> Thanks!
> 
>  
> 
> Richard.
> 
>  
> 
>  
> 
>