[OpenAFS] Why does CopyOnWrite corrupt RW volume during non backup times

[Loni Ly]

Hi,

Based on the listing below it appears that COW occurs
when cloning does.  However, I have a situation with a
"GYM" user who has had this happen several times
during backup and regular hours.  The RW is a home
volume without a RO site, just a backup volume.  The
backup volume has been zapped several times to resolve
issues, but always reappears with the same volume ID. 
How can I remove the backup volume completely and
recreate it with a new volume ID?  Does this require
some changes to the RW volume?

COW only hits this user in the cell.  Be great to
quiet this permanently, but I'm at a lost for now.
Thanks in advance.
BTW: Cell is 1.2.13...more info needed?

https://lists.openafs.org/pipermail/openafs-info/2002-June/005013.html

On Wed, 26 Jun 2002, FBO wrote:

> Hello,
> 
> what is CopyOnWrite? What does it do?

When you clone you get pointers to the contents on
disk as they exist. If
you write something you get a CopyOnWrite as the "old
volume" gets the
original contents and the "new volume" gets the new
contents

> > ROs are clones just like the backup volumes so
this won't help.
> As far as I understood Cow is a function, copying
the differences of
> a changing RW- to a Backup- or RO-VOlume.

> Shouldn't the Bug be avoided if RO-Volumes are used
> and I have them them on a Partition different to the
RW's one
> (with the disadvantage of needing more diskspace for
ROs)?

you still clone a volume to send a copy when you
release, so, unless you
don't release it...



		
__________________________________ 
Yahoo! Messenger 
Show us what our next emoticon should look like. Join the fun. 
http://www.advision.webevents.yahoo.com/emoticontest