> > Yes, there is, but you run into data integrity problems, and data
> > synchronization problems.   ...

My experience has been different.
But I understand the point and agree that there is a large exposure.
Experience from read-only volumes mounted read-write and using the
older DASD driver would seem to suggest it can get NASTY.

 ...
> > possibility would be to have multiple copies of /usr, and update the VM
> > guest's CP directories to point to the most recent one so that when the
> > system is shutdown, logged off, and back on again, it will pick up the new
> > file system.

Or use a DCSS in place of DASD.   ;-)

> >        Unless you're talking about enough images being run that DASD
> > space is still conserved, it's kind of pointless.

It's not exclusively a resource conservation thing.
It's also a system management thing.

Say you're migrating from Wintel to L/390.
If all you do is put N PC workloads into N v-machines
you haven't reduced any of the manpower load or headache,
you've blown a good portion of what we could win with zSeries.

Correct:  three images isn't enough to bother sharing.
Correlation:  three images isn't enough to justify zSeries.

> Could someone with /usr mounted ro try this?

Try which part?   Writing to a volume that is read-only to others?
Been there;  done that;  and it did not cause me any pain.
I did NOT run any exhaustive test,  so  "it depends".   ;-)

> mount /usr -o remount,ro

A good thing to do is mount the shared volumes read-only on the master
and then  'mount /usr -o remount,rw'  when you need to update.
Hopefully you won't need to do this more than once or twice
before re-IPLing the sharing systems so that they'll be current.

Reply via email to