[mythtv-users] 30 minute freezups
Kevin Kuphal
kkuphal at gmail.com
Fri Dec 19 18:42:55 UTC 2008
On Fri, Dec 19, 2008 at 12:37 PM, Greg Woods <greg at gregandeva.net> wrote:
> On Fri, 2008-12-19 at 11:50 -0600, Carl Fongheiser wrote:
>
> > I think you're tilting at windmills here. At least where I live, it
> > costs about 1 cent a day to let a hard drive spin at idle.
>
> Power cost isn't really the major issue. I'm more worried about drive
> lifetime. Even the cost of a new drive isn't the major problem, it's the
> major hassle involved in pulling the system out of the rack,
> disconnecting and reconnecting all the cables, opening it up, and
> replacing a drive, plus losing those recordings. My time to do that is
> worth way more than a year of electricity to run a hard drive. I know
> that spinning up and down is more stressful per second than just letting
> it run, but the truth is that my Myth system is only in use perhaps 4-5
> hours a day on average, including recordings (more on weekends when
> there are football games, less during the week when the system might be
> idle for several days). I have not, as you say, "done the math", but I
> have to believe that I'm better off having the drives idle for 19-20
> hours a day than having them run all the time. Heat generation is also a
> factor, especially in the summer.
You should look at the research done by google on drive failures and see
that what you're doing probably isn't helping this at all
http://research.google.com/archive/disk_failures.pdf
They pretty much concluded that temp and activity had very little to do with
failure rates. If you're concerned with drive replacement/time/loss, you
should consider a RAID 5 option as this will allow you to recover in the
event of a single drive failure without data loss and most of the time,
without downtime.
Kevin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-users/attachments/20081219/392c87a6/attachment.htm
More information about the mythtv-users
mailing list