[Info-vax] Request description of UFS for VMS person
Bob Koehler
koehler at eisner.nospam.encompasserve.org
Tue Apr 21 09:19:23 EDT 2009
In article <33c33267-2172-433d-9a98-fdd1b9cbda3b at y7g2000yqa.googlegroups.com>, AEF <spamsink2001 at yahoo.com> writes:
>
> Then there's the mysterious VMS deletion process which I'd love to
> learn. I only know about it from the strange ANAL/DISK errors about
> file headers being busy, deleted, quasi-deleted, or whatever. I'd love
> to learn about that too.
This is an area in which every file system I've studied operates the
same. There is no file delete primitive at the lowest levels of
the VMS or UNIX kernel, and I think not Windows, MS-DOS, or classis
MacOS, either.
To delete a file the atomic operations are: open the file, set the
delete-on-close flag, and close the file.
Every OS has problems if you try to delete a file that something else
has open as that last step can't complete until all things close the
file. How this is dealt with varies, somehow the OS has to keep
track that the file is really there and somehow provide a way to
finish the delete. The actual last close may not happen until system
shutdown, and in VMS sometimes it actually just doesn't happen during
system shutdown. It is possible to always clean up during boot or
mount, but on VMS you get to do it manually at your convinience.
I generally have the automated backup scripts rebuilt the bitmap
once a week before doing full backups and I find the delete in
progress files so rare that unless I'm doing a product install I
do it manually every other month or so.
More information about the Info-vax
mailing list