For the past month or so, my SATA RAID-1 configuration of two WD 36.7GB Raptors (about 3.5 years old, and via the WinXP Promise software RAID controller) has been experiencing progressively more bizarre performance.
One or both of the drives (I can't tell) seems to spin down and shut off (complete with *click-whirr-
whirr* sound) after 2-10 minutes of no access. The other oddity is sort of a corollary: when accessing
some sections of the disk, the read time is incredibly slow -- probably around 15MB/s. While reading these sections, a *click-whirrrr* sort of spin-up sound can be heard every 6-10 seconds. And I emphasized "some" because other sections read just fine, even just moments after the aforementioned apparent spin-down/shut-off. On top of that, the slow sections sometimes read normally, seemingly for no reason at all. However, over the course of the last month, and especially the last week, those slow sections seem to be becoming more common.
I've run chkdsk, and it detects no bad sectors on either drive. I couldn't find any other diagnostics to run, because most of them seem to be designed to work on normal, non-RAID drive configurations, and the remainder are software written exclusively for specific hardware RAID controllers. I can't even find a simple S.M.A.R.T.-value reader. Western Digital's own Data Lifeguard Tools doesn't even recognize the drives as S.M.A.R.T.-capable, at least not from within the Windows version of the tool.
So, does this sound like an impending drive failure? How can I be sure when there are no diagnostics to run?
Posts
It's still kinda voodoo. Google did a white paper on hard drives a while back and they basically found S.M.A.R.T. to be next to useless in determining hard drive failures. If the drives feel like they are acting up, then replace them.