smart(8) Call for Testing
Michael Dexter
editor at callfortesting.org
Tue Mar 27 19:39:46 UTC 2018
On 3/27/18 10:01 AM, Charles Sprickman via freebsd-fs wrote:
> Again, maybe I’m just missing something or maybe this is here for a particular vendor that needs it or something.
You are all welcome to watch the AsiaBSDCon and BSDCan talks on the
rationale but here are a few:
1. Ability to use it in-base. Chuck has done the "hard part" and more
can be built on top of it. With an in-base utility or library, the
installer could check disk health before installation as could zpool(8)
(proposed elsewhere with the optimal implementation TBD).
2. Scriptability as per my last post. Most administrators have
relatively homogeneous selections of disks and seek four to five key
SMART values.
3. Usability. You CAN obtain even raw-er values with a camcontrol
inquiry but that truly is a pain and as I recall, they land in hex
values, making scripting even more a pain.
For what it's worth, I believe smart(8) had NVMe support before
smartmontools did.
smart(8) is not the work of a vendor but rather is based on my personal
experience with hundreds of ZFS-based storage systems in the wild.
I made several public calls for input on the design prior to Chuck
writing any code. May I kindly request that you rephraise your
criticisms as feature requests?
Michael
More information about the freebsd-fs
mailing list