On 09-09-30 11:24:22, Richard Heck wrote:
I am getting a lot of kernel errors of that type:
hde: lost interrupt
on one of my servers. (This is actually CentOS, not Fedora.) The
errors are intermittent, but when they start they often will just
keep coming, and the disk has locked up a couple times completely.
Something presumably needs to be replaced, but the question is: What?
The drive is a 500GB EIDE drive, connected via a Promise Ultra 100TX2
controller to some ancient motherboard. (This is a 300MHz Pentium II,
acting as a server, including a media server.) The controller is
needed because the old mobo bios won't deal with such large drives.
There's another drive connected to the Promise, as hdf, and I'm not
seeing errors from there. So, all in all, it looks as if it's
probably the drive, even though the drive is less than a year old.
But before I replace it, I thought I'd ask and see if anyone has any
other ideas. It'll be a hassle to replace it, since it's part of a
volume group, etc.
Cable? Either a better cable, or improve its routing, or just crumple
it so it can't cozy up to anything else, or simply unplug and replug
it? Also, what's on the power connector? Unplug and replug that one
too.
Replacing the drive can be done by adding a new drive to the Volume
Group and using pvmove (not that I've done it).
--
____________________________________________________________________
TonyN.:' <mailto:tonynelson@georgeanelson.com>
' <
http://www.georgeanelson.com/>