[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Edlug Archive Mar 2004 ]

Re: [edlug] Failing hdd and mystery crash

On Thursday, March 4, 2004, at 11:34 pm, Peter G. Hancock wrote:

Mike Moran wrote (on Thu, 04 Mar 2004 at 18:43):

Ok doke, I've done the whole badblocks thing, and lo-and-behold it
found 10 bad blocks. However, when I try to get e2fsck to fix them
(from within a knoppix boot), using the "-c" option, the final output
says "0 bad blocks", amongst the normal summary. What gives?

I think what gives is that the badblocks program is looking at a _partition_ (i.e. approx an array of blocks), whereas e2fsck is looking, a bit superficially, at a _filesystem_ represented in parts of that block array, checking that some of those blocks look like coherent directory/file inodes and w.h.y. . The bad blocks could be marked unallocated, or something.

That doesn't make any sense; after all e2fsck just calls "badblocks" itself. As far as I am aware the only controlling arguments it passes are the block size. Could it be possible that 10 badblocks show up at the default "badblocks" blocksize, but not at the blocksize supplied when run by e2fsck?

So, badblocks is giving you something closer to the real deal about
the state of your disk.  The file-system may only be running into them
occassionally.  Those that it sees are maybe recorded in inode 2, but
it's not the real scoop.

Roy Bamford suggested "man smartd/smartctl".  He's quite right.  If you
find and install the smartd demon, and your disk is not too ancient, it
can be persuaded to tell you more than you ever wanted to know about
your disk.  You'll never sleep soundly again...

I have installed smartd, and that's another part of the story (which I'll post about later when I'm at the machine). I've also still to compile the other tool that was mentioned.

Mike http://www.houseofmoran.com/blog/

You can find the EdLUG mailing list FAQ list at:

This archive is kept by wibble@morpheux.org.DONTSPAMME