On Tuesday, 03/10/2009 at 12:19 EDT, "Edmund R. MacKenty"
<[email protected]> wrote:

> Yes, but you *really* don't want to do that.  Your guest with the RO
minidisk
> will get corrupted data.  You see, Linux caches blocks it has read from
the
> filesystem in memory.  So imagine that it reads in a block containing a
set
> of directory entries and caches that.  Now imagine that another guest
with RW
> access to that filesystem removes that directory.  The RO guest won't
know
> about it: it will still happily use that cached directory block when
reading
> that directory, which contains references to files that no longer exist.
> What happens when it tries to read those files?  It reads those blocks
from
> DASD, which may well have been overwritten by the RW guest with some
other
> data, because those blocks were freed up when the directory was deleted.
> Oops!

For large-scale sharing, xip2fs and DCSSes are the only way (IMO) to share
files in the way we all originally envisioned minidisk sharing.  The DCSS
owner makes all the updates and then saves a new copy of the DCSS.  New
users get the new one - previous users still have the old DCSS.  When no
one is using the old one any more, it magically disappears.  And the
DCSSes always have consistent content, even if someone loads the DCSS
while you're in the middle of building a new one.

Without unionfs or its moral equivalent, it requires a clear separation of
config, runtime temp storage, and executables into separate directories.

Alan Altmark
z/VM Development
IBM Endicott

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to