Wiki‎ > ‎

White Label hard drives

posted Jun 3, 2016, 10:17 AM by Dong Xu   [ updated Jun 24, 2016, 3:04 PM ]

We shoved all the drives in 36 bay Supermicro chassis and LSI 9280 cards and built out 3x12 disk RAID-0 configurations. From there, we would monitor the LSI card for errors as well as the system logs. For testing, we ran both a patrol scan on all the disks as well as a ---- badblocks -p 100 -w -v /dev/sdb ---- just to make the system do lots of reading and writing. If you have other good testing methodologies, feel free to tell me.

Drive burn-in script attached below and instructions:

This process also prompted me to start documenting the tools that I am using and building more. They are available at: chuckleb/linuxdisktools · GitHub. It's a bit barren right now but I'll be writing up some in the next few weeks.

A lot of us use these drives and have discussed all the different types of drives being shipped under the same generic white label. What we've learned is that there is no guarantee what drive will be delivered; you might have received some from a bad batch. If you choose to use these drives for important data (like I did), it is best to use a resilient file system like ZFS and RAIDz2 or RAIDz3 and couple that with ECC RAM. RAIDz1/2/3 protects against silent corruption and FreeNAS gives you the option of running a scheduled scrub (verify) against the array to identify if you have a drive that is starting to go bad.

Personally, I will keep using these drives and purchase more if I have a need because I am confident in the ability of ZFS's RAIDz3 to identify and overcome any issues with the drives

Monitoring Hard Disks with SMART

Dong Xu,
Jun 3, 2016, 10:43 AM