Re: 2.1.56-SMP: still no problems for me
Brian Blackmore (bnb@looking-glass.org)
Sun, 21 Sep 1997 14:20:33 +0100
H. Peter Anvin wrote:
>
> > > > > I can do nothing to kill these processes. The NFS mounted FS
> is indeed
> > > > > unmounted, and there is no zombie umount process. Using top
> gives no
> > > > > indication of who is chewing the CPU. A compute-bound process
> manages
> > > > > to get over 98% of the CPU, so it doesn't appear that the
> phantom is
> > > > > using much CPU.
> > >
> > > Load average != CPU usage, especially on Linux. A D process under
> > > Linux is not using the CPU, but counts toward the load average.
> >
> > Yes, I know. Nevertheless, it doesn't seem right. Especially when
> each
> > new mount bumps up the load average thanks to a new lockd process.
> > Anyway, this odd behaviour may be connected with the autofs
> > problems. Then again, maybe not...
>
> This is certainly broken, but unless it only happens when mounting
> from autofs, which doesn't make sense to me, it sounds like just
> simply a knfsd problem unrelated to autofs.
This is probably a completely bogus suggestion, but what the hell :-)
When autofs mounts a NFS mount, there is a process that immediately
wants to access said mount. On a normal mount this is not the case.
Somebody said they managed to trigger the same problem via an umount. So
could we be looking for some way that a process manages to access an
mount when it is actually unmounted?