On Fri, Oct 14, 2011 at 3:46 PM, <[1]sf...@users.sourceforge.net> wrote:
Ken Trumbo:
> Yes, I don't want /aufs/.wh.fileA. If the user removes the file
/aufs/fileA
> then I just want the file /rw/fileA to be deleted (which I assume happens)
> and the original file /ro/fileA will show through.
Hmm, so you want such like this?
- both of /ro/fileA and /rw/fileA exist.
- /aufs/fileA exists too.
$ rm /aufs/fileA
$ cat /aufs/fileA
and "cat" shows /ro/fileA.
If so, that is not normal fs behaviour, and I don't know why you want
such behaviour.
Yes, I know it sounds a bit strange at first. For example, the user can
modify a file and test the change (say to a file in /etc) and if they don't
like the change deleting just reverts back to the previous version.
Instead of aufs, you may want try "mount --bind fileA" and then unmount
it. It may be possible if you develop something based on FUSE. But I
don't think it is a good idea.
I can investigate FUSE but I think aufs is very close to what I need. I am
wondering if it is easy to patch aufs to not create the whiteout file. Or,
can I put the whiteout file into some 'nullfs'.
References
1. mailto:sf...@users.sourceforge.net
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2d-oct
