Re: Hard Lock-Up Problems
"James M . Mastros" wrote:
> 1) Are their any unusual noizes noticable either during or slightly before
> this lockup?
Not unusual as such, like say the sound of a HD crashing (been through
that) but there is a reproducible pattern. There is usually a heavy
disk load (possibly this just makes it more likely to occur), then it
stops. A second later there's a split-second disk access, then another
second later it hangs.
> 2) Is your hard-drive under warranty?
No. That sounds ominous.
I've discovered that if I keep the disk space over 50Mb (100 to be
sure), files don't seem to be written where this stuffs up.
> 3) Do you get any log messages before lockup?
No.
> 4) Will pressing numlock toggle the LED?
No.
> 5) Does shift-scrolllock give you less then 128k of free swap?
When it's crashed shift, alt and ctrl scrolllock don't work. Before,
fsck uses up all memory but never starts using up swap. This is what I
would expect from something that reads the whole disk through, it's just
caching what it can, but not actually requiring the memory.
> 6) Have you run e2fsck -c -C0 -vv <root device>? Have you run it repetedly?
> ('man e2fsck' before running this command. Make shure you remount root
> readonly. Having messed that up once, let me assure you it is not a good
> thing.)
I just did that. It seems it found a bad block. The following was the
output;
hda: read_intr: status=0x59 { DriveReady SeekComplete DataRequestError }
hda: read_intr: status=0x40 { UncorrectableError }, LBArect=3646226
sector = 1327826
end_request: I/O error, dev 03:04 (hda) sector 1327826
That's as good as I can transcribe it.
I ran it multiple times. The second time and later the error came out
twice but I think it only came out once the first time. Anyway, this
always crashes on "checking directory structure" in the usual way.
I ran badblocks -o and discovered the bad block was "663913". Running
multiple times seems to just come up with this one bad block. This is
not where I would have suspected to find it, more like 770xxx. Even
though e2fsck man says it marks the blocks it finds as bad, it still
seems to crash e2fsck.
Thanks for your help.
--
Matthew Tuck - Software Developer & All-Round Nice Guy
mailto:matty@box.net.au (ICQ #8125618)
Check out the Ultra programming language project!
http://www.box.net.au/~matty/ultra/
Reply to: