Meantime to failure on HP 18gig internal drives

[Jason Berendsen]

Hello,

We have had 3 18gig internal disk drives go bad on our N boxes in the last 6 weeks. These failures are occurring on different boxes. These systems are almost 2 years old. Does this seem like a large amount of failures for such a new server? Also, does anyone have any stats on what HP states the meantime to failure is on these drives.

[Patrick Wallek · Accepted solution]

The drives you have are probably rebadged Seagate drives. I have a bunch of 18GB drives in Jamaica enclosures and most, ir not all of them, are Seagate ST318275LC drives.

Now according to Seagtes website, the MTBF for these drives is 1,000,000 hours.

See this web page for more info:
http://www.seagate.com/support/disc/specs/scsi/st318275lc.html

I would not be surprised, though, to have some drives start going after a couple of years, especially if the drives are heavily utilized and don't sit idle much.

[Sridhar Bhaskarla]

Hi,

We have N-Servers here for more than 2 years and no problems so far. Do a diskinfo /dev/rdsk/c#t#d# and check out the model. If it is cheetah 18XL (ST336704LC) then here is the link to it's specifications and reliability.

http://www.impediment.com/seagate/s2000/spec_318404lc.shtml

-Sri

[Jeff Machols]

Jason,

The odd thing here is that all of a sudden this is happening. You might want to focus on environment issues. Usually heat or acidic air will affect memory first, but you never know. maybe it's time to replace air filters and check inside the boxes for excessive dust.

[harry d brown jr]

Patrick, 1,000,000 hours is about 114 years! Are they insane, or is that if they stay shrink wrapped on the shelf :-????

Jason, I hope you still have maintenance on the systems.


live free or die
harry

[A. Clay Stephenson]

Hi Jason:

Patrick's numbers are approximately correct depending upon your exact model. Typical values are somewhere between 700,000 to 1,100,000 hours. Failure rates in UNIX busy boxes tend to be higher because the duty cycle for the drives is mush higher. Also, heat is a major contributor to drive failure. Make sure that your HVAC system is adequate for the thermal load and that airflow around your drive enclosures is adequate. Just as a reference point, I have a little over 150 drives in various enclosures and it is not unusual to replace a drive every month or two although somestimes 3 or 4 in a week. The main thing is to anticipate this and have all drives mirrored (or RAID'ed) so that this is not a big deal. Hopefully, all of your drive are hot-plug or hot-swap so that no downtime is needed. In my case, I even keep spares on hand.

[Patrick Wallek]

Harry,

I agree that the 1,000,000 hours sounds a bit ridiculous. They also say that that is power-on hours. So maybe if you just power the drive up so it is spinning, but never actually use it so the heads never move and it is never stressed......