Re: Infamous 'root busy' on shutdown <=> unkillable processes on wait IO

Andreas Bombe (andreas.bombe@munich.netsurf.de)
Sat, 30 Oct 1999 01:33:34 +0200

On Thu, Oct 28, 1999 at 10:36:22PM +0200, François Désarménien wrote:
> Since 2.2, there often happen to have processes in an 'unkillabe' state, waiting
> for I/O usually on Unix sockets. This I/O never comes, so the root FS will be busy
> forever.
>
> Why don't those processes have a slice when they get the KILL signal, so they could
> cancel the IO and gracefully exit ? I though that when you do a read and a signal
> (caught or not) is receive, the I/O was cancelled and the read returns EINTR.

Only when the kernel code in question does interruptible sleeps. This
seems not to be true in your case and the code deadlocked somewhere.
You apparantly encountered a kernel bug.

Try again with a more informative bug report (which processes are
unkillable, in which function do they sleep ('ps axl', WCHAN field)).
The usual stuff.

> This is really annoying, the only turn around I found is to sync the filesystems
> just before the final shutdown step, but if it preserves data integrity, it still
> takes the (looooooooong) e2fsck time at reboot.

You can avoid the fsck when you remount the filesystem read-only. You
can do this either with 'mount yourfs -o remount,ro' or 'umount -a -r'.

-- 
    Andreas E. Bombe <andreas.bombe@munich.netsurf.de>
    http://home.pages.de/~andreas.bombe/
RSA 0x886663c9  30 EC 09 73 84 7B 55 83  C4 7A 91 D9 9D C5 4B B0
DSA 0x04880A44 72E5 7031 4414 2EB6 F6B4  4CBD 1181 7032 0488 0A44


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/