Badblocks test for WD60EFRX?

[TooMuchData]

I'm burning-in some 6TB WD Red drives. I attempted to run badblocks and got the following error:

[jeff@freenas ~]# badblocks -wsv /dev/ada0
badblocks: Value too large to be stored in data type invalid end block (58605225
84): must be 32-bit value

The command fails for any of the 6TB drives, but runs properly for a 1TB drive. Does badblocks not support the larger disks?

Any suggestions will be appreciated.

Running FreeNAS 9.3 Stable 20150109 on ASRock C2750D4I with Crucial (2 x 8GB) CT2KIT102472BD160B, Seasonic SSR-450RM, Node 304 case. No Zpools defined yet.

Thanks for your help.

 

[Ericloewe]

If I'm reading that right, badblocks is limited to 2^32 sectors...? Since it defaults to 1024 byte sectors, it would seem that badblocks can only be used on disks up to 4TB. That's... very weird.

For now, I'd recommend you try jgreco's script instead.

Edit: Setting the block size to 4096 might allow you to hack around the problem.

 

[Fraoch]

Interesting, I looked it up.

Support won't be added:

http://unix.stackexchange.com/quest...bit-integer-as-start-end-values/152183#152183

:(

But there is a workaround if you increase the block size?

http://askubuntu.com/questions/548945/how-to-check-badsector-on-ext4-6tb/552343#552343

 

[TooMuchData]

Thank you Ericloewe and Fraoch.

Increasing blocksize to 4096 results in an error "Inappropriate ioctl for device", but the task begins and continues.

I'm set for now. :)

 

[Ericloewe]

Doesn't matter. The drive uses 4k sectors internally, anyway.

 

[TooMuchData]

Thanks again, Ericloewe.

I'll ask another question, please.

My new WD 6TB drives average 170 MB/s serial and parallel read rates. The individual rates range from 165 (97%) to 173 (102%). Does that variability seem reasonable? (I would think so, but don't know how consistent these devices should be). None of the drives have thrown an error in 10 days of constant testing, and none of them has any reassigned sectors.

 

[Ericloewe]

I don't have hard data to offer you, but +-3% is not something I'd worry about.

The only problematic ones are the obviously bad ones, where no question remains.