Re: can I mark sectors as bad?

[Russell Neches <russell@ccs.neu.edu>]

Check the to see how hot the drive is getting. If it burns your
fingers (or your friend's fingers), it's quite possible that the drive
is seriously overheating. This can cause an assortment of bad things
to happen. It's possible that the bearings seize up as they
expand. The armature could be going out of alignment and confusing the
drive logic. CMOS chips have a tendency to glitch and eventually stop
working at high temperatures. Or, the drive could be powering down to
save itself from being cooked. 

Check the ventilation in the system and the ambient temperature of the
room. Just use your judgment - if it's an elderly drive, the operating
specifications are probably optimistic at this point. If you cobbled
the system together in a hurry, you might have some airflow problems
with ribbon cables.

If you're lucky, that will fix (or at least assuage) the problem. You
should get a new drive anyway, though. I keep a pile of 3 GB IDE hard
drives around just in case - I'm sure you could obtain one easily
enough.

Russell


"Captain Weenie" <weenie@cybercede.net> writes:

> Anyone familiar with the workings of the file system:
> 
>  I assume the hard drive is flaking out - but the error is mostly the same - a
> power cycle gets things running again without error.
> 
> I am getting the following error -
> 
> wdc0:0:0: not ready, st=0xfe, err=0x00
> wdc0 channel 0: reset failed for drive 0
> wd0a: device timeout writing fsbn 968608 of 968608-968623 (wd0 bn 968671; cn
> 960 tn 15 sn 46), retrying
> 
> ...
> and so it goes but the st=0xfe, err=0x00
> changes to
> st=0xde, err=0x00
> 
> ,,,
> 
> and this just repeats over and over - system seems completely locked up - I
> pull the plug (ouch) - give her juice again - and she complains momentarily
> about the filesystem not being properly unmounted - but then comes up clean.
> 
> while performing an fsck -f -v (and several other times on its own) the wd0a
> times out - but then recovers reporting soft error (corrected)
> that error most recently looked like this:
> 
> wd0(wdc0:0:0): timeout
>             type: ata
>             c_bcount: 8192
>             c_skip: 0
> wd0a:  device timeout reading fsbn 750400 of 750400-750415 (wd0 bn 750463; cn
> 744 tn 8 sn 7), retrying
> wd0: soft error(corrected)
> 
> the fsck does not show any troubles - unless 639 frags is an issue.
> 
> this was a system that I threw up quickly to get a job done - so the entire
> filesystem is mounted on /
> I do usually break out more mount points - I don't know if that could be an
> issue here or not.
> 
> I haven't rebuilt the kernel on this - so there hasn't been an opportunity for
> any bad magic.
> 
> I plan on getting some other boxen up to replace this one - but I'm hoping
> there is something I can do to prolong the life of this beast of burden.
> 
> Any way to mark sectors as off limits or something?
> 
> I loaded this system on that fateful night at the end of last month - it was
> late night on the 30th that I pulled this system's OS down off of the Canadian
> FTP server - I'll be posting a question about anoncvs on the misc list if
> anyone is interested in following.  I felt this issue was of a more technical
> nature - and more suiting for this list.
> 
> I haven't posted to this list before - so if this is OT please use a lighter
> as opposed to a flame thrower.
> 
>     Danke,
> 
>             Capt. Weenie