This made the errors on boot and shutdown go away, however, I'm still not able to
go to runlevel one and un-mount / and then do a raidstop.  It doens't complain
about:
umount /dev/md0
but if I type mount immediately after this I still get /dev/md0 mounted to / in
read-write mode.  I've checked and mtab says this too.  However, if I try to edit
the mtab manually (delete the /dev/md0 entry) it says that mtab is read-only.  I
try to force it with a w! and it says that the file system is readonly.  Why is it
still showing up as mounted if it is in fact not mounted?

Of course if I reboot, start out at a clean state, go to runlevel 1, delete the
entry from mtab, and then umount the device everything shows up correctly except
if I try a raidstop it gives me an error of:
/dev/md0: Device or resource busy
which means that it thinks that since we are still sitting at the / prompt that
the device is busy (I believe).  In theory since the device is "umount"ed
shouldn't / be a readonly partition in RAM?

Anyway, as / and everything under it are in md0 (my boss is a total and complete
dolt sometimes) we are having trouble convincing it to stop.

Chris

"m. allan noah" wrote:

> if using the default redhat patched kernel,
>
> edit the file /etc/rc.d/init.d/halt, and remove the section that does the raid
> stop starting around line 89.
>
> you will need to make sure that you have enabled autodetection and that the
> raid personality that you need is included either directly into your kernel or
> is in your initrd of course. you will also need to make sure you are running
> the .90 raid code (redhat default) and your partitions are marked as type fd.
>
> allan
>
> Chris Tooley <[EMAIL PROTECTED]> said:
>
> > I have a problem that I am having a hard time figuring out.
> >
> > I just set up a new server with Red Hat 6.2, with 2 20GB IDE ata-66 drives.
> > I partitioned each drive into two partitions one swap of 150MB and the
> > second into 9630MB. I configured /dev/md0 for / to boot of the 9630MB mirror
> > and /dev/md1 for the swap mirror.
> >
> > my problem is that when I shutdown the box I get the message:
> >
> > ****************************************************************************
> > *******
> > Turning off Swap                                                [ OK ]
> > Turning off quotas                                      [ OK ]
> > md_do_sync() got signal ... exiting                     [ OK ]
> > Turning off RAID for md0 md: md0 still mounted.
> > /dev/md0: Device or resource busy
> >                                                         [FAILED]
> > Turning off RAID for md1 interrupting MD-thread pid 7
> >   raid1d(7) flushing signals.
> > marking sb clean...
> > md: updating md1 RAID superblock on device
> > hdb5 [events: 00000016](write) hdb5's sb offset: 160512
> > hdb5 [events: 00000016](write) hdb5's sb offset: 160512
> > .
> > unbind<hdb5,1>
> > export_rdev(hdb5)
> > unbind<hdb5,1>
> > export_rdev(hdb5)
> > md1 stopped
> >                                                         [ OK ]
> > Unmounting proc file system                             [ OK ]
> > Please standby while rebooting the system...
> > ****************************************************************************
> > ********
> > The system then reboots and has to rebuild the sb struct offset on md0.
> >
> > I read through your HOWTO to try and find something I am doing wrong but
> > couldn't find anything on this topic. Then tried to do # raidstop /dev/md0
> > and get:
> >
> > Turning off RAID for md0 md: md0 still mounted.
> > /dev/md0: Device or resource busy
> >
> > If I go to init 1, then try to umount /dev/md0, it still tells me the device
> > or resource is busy.
> > Do you have any ideas on how to rectify this situation?
> >
> > Thanks for your time.
> >
> > Chris Tooley
> > Joslyn Art Museum
> > http://www.joslyn.org
> >
> >
> >
>
> --

Reply via email to