Re: getting Linux into business..

Stig (stig@hackvan.com)
Wed, 21 Aug 96 07:43 PDT


Eric Youngdale wrote:
>
> I am wondering whether umount should be made more robust so that
> it essentially forks prior to each unmount attempt. When all the children
> have died we know that everything is unmounted, and we can proceed. If
> one of the umounts hangs, then at least all of the remaining partitions
> are cleanly unmounted.

Sounds like a really good start. Maybe it's even 90% sufficient.

A related problem is that other hung processes will have their current
directories marked as active and/or open. Those processes won't die and so
present an obstacle to unmounting other partitions besides the hung one.
Typically these hung processes will be something along the lines of df(1).

> If some NFS partition cannot be unmounted, can we still force the
> root partition back to R/O status so that it is also marked clean?

There might also be a problem with a scsi device that's disappeared, perhaps
because it was generating errors or perhaps because it was turned off, or...

I would really like to see hot swap capability, too, by the way. I'd like
to be able to move a tape drive between several machines w/o rebooting.
Since they're not all unix, network backups aren't an option.

Stig