better method for weeding out bad drives

[digity]

I picked up a ton of used SAS drives (3.5" & 2.5") for free with 3 to 8 enterprise years on them knowing I'll have bad ones. Some won't spin up at all and some spin but won't mount. The remaining I've thrown in stripes (FreeNAS pool) and have done poor man stress tests by transferring large sets of data back and forth with rsync and cp. Initially this weeded out a good amount of failing drives (SMART failures), but with each advancing round they're failing less and less. Obviously this method is slow and tedious - is there a faster method to weed out bad drives??

 

[Matt_G]

Run a long SMART test.
If that passes, then run badblocks and then another long SMART test.
Obviously, in the commands below xxx is the device name in FreeNAS.
I run these from an SSH session using TMUX so I can do multiple drives at a time.

• smartctl -t long /dev/xxx
• badblocks -b 4096 -wsv /dev/xxx

It's not fast, but it is hands off and thorough, and you can do as many drives at a time as you have connections.