Dear Antonio,

> For an errorless, buffered copy (not direct access) like the one you are
> doing, you may try increasing the --cluster-size to 256 or 512, but most

Setting the cluster size to 256 improved things a bit, but it was 512 that 
really made a difference. I'm even trying with 1024 (hope it is safe), and 
it's also a bit better but not by a lot.

> probably the bottleneck is in the sustained writing speed of the destination
> partition.
I assume so too, but when I did the first ddrescue copy (/dev/sdb1 to 
sdb1.img), it was an almost identical config, except the original machine
was even older!

> 
> At 250 MB/s your drive seems faster than mine (25 MB/s), but not fast enough
> if your partition has 500 TB (12 days x 500 MB/s).
> 
Well, that's quite close to our scenario, copying RAID to RAID over a 10G 
connection. That's why it's a bit scary to let it run so many days and
then find out some setting was wrong, but your reply was reassuring :)

> 
> Best regards,
> Antonio.

Thanks a lot and all the best,
Alex



Reply via email to