Civileme wrote:
>
> On Sunday 06 May 2001 10:20, Benjamin Sher wrote:
> > Dear friends:
> >
> > [Using Lm 8.0]
> >
> > Would appreciate your help with the following problem:
> >
> > Sometimes LM 8.0 will reboot properly and sometimes it won't. And
> > sometimes it will shutdown properly and sometimes it won't. When it
> > doesn't, Linux freezes at "Please wait while Linux reboots" or at the
> > very end of the "shutdown -h now" sequence at "unmounting file systems"
> > instead of "System halted" or "Power down." These two problems may be
> > connected. This is more than just a nuisance. When LM 8.0 does not
> > reboot or shut down properly, it goes through its "not clearly
> > unmounted, check forced" routine. Of course, with ReiserFS, this now
> > takes only a second instead of 10 minutes. But, still, it feels me with
> > anxiety and worry that there is something really wrong.
> >
> > NOTE:
> >
> > Whenever I log on, I get the following strange message in the console:
> >
> > sher07 login: Winbond Super-IO detection now testing ports 3F0, 370,
> > 250, 4E,
> > 2E, SMSC Super -IO detection now testing Ports 2F0, 370.
> >
> > I then have to type CntRL-C to return to my usual logon screen with the
> > prompt, so I can log INTO KDE with startx.
> >
> > Any idea if this has anything to do with the reboot and shutdown
> > problem? Could all of these issues be related in some way?
> >
> >
> > Thanks so much.
> >
> > Benjamin
>
> OK some daemon isn't stopping and is using a filesystem.
FWIW, I've seen this too... however, in my case, it's usually NFS mounted
partitions that cause the problem. Shutting down with an NFS mount will
sometimes cause one of the server's nfsd to crash. I've posted the crash (seen
on LM6.1 & 7.2) on the cooker list (I can resend if you want) leaving some data
on the RxQ making all the other nfsd useless and requiring a reboot to fix.
IMO, shutdown should unmount non-local drives before proceeding... I haven't
checked the source; but I wouldn't be surprised if NFS does not handle a
shutdown properly if at all...
HTH,
Pierre
> Using the message you are giving me, I would do this
>
> as root in console or konsole
>
> # rgrep -i -l -r "unmounting file" /etc/rc.d
>
> And find out which initscripts on shutdown are making that message
>
> Then edit the initscript so you have a chance to see which filesystem it is
> hanging on.
>
> Or, alternatively, list out the mount points as in this example added line,
> right before the unmount.
>
> fuser -k -9 /home -k -9 /var -k -9 /usr -k -9 /
>
> This will bring down all the processes using any of those filesystems, hard.
> It is something of an extreme solution, but it may be necessary for a daemon.
> More likely, it is easier to solve. A daemon needing updating or
> reconfiguration is probably using /var and it should be possible to update or
> reconfigure the daemon to stop gracefully. fuser is also good for listing
> processes using files or sockets--man fuser for details.
>
> Civileme
--
Support Linux development: http://www.linux-mandrake.com/donations/
