[OpenAFS] verifying ro volumes are not taking up extra space

[Todd_DeSantis@transarc.com]

Hi -

You can do several things to check on the paritition.

You should have a tool called 

	volinfo

in /usr/afs/bin, or it can be built in the src/vol directory.

Yo can check space quickly via  (run as root on the fileserver machine)

 - /usr/afs/bin/volinfo -sizeOnly

   and it can report on the sizes of your volume.  Make sure that you
   don't have any strange looking volume names like

	move.temp  that means a failed "vos move", etc
	The above name might not be exact, but look at any
	strange names like this.

 - You can also get a list of the vnodes/inodes used by each volume
   and you could do this

   /usr/afs/bin/volinfo -part /vicepX -volum <RW numberic id> \
	-vnode -inode > volinfo.<volume>

   /usr/afs/bin/volinfo -part /vicepX -volum <RO numberic id> \
        -vnode -inode > volinfo.<volume>

   And then you can compare the inodes that are listed for the files/
   directories to make sure that they are pointing to the same inode.
   If you have a recently released ReadOnly volume, you should
   probably see that the inodes are the same indicating that the RO is
   in fact a clone of the RW.

 - When we saw really bad discrepancies between like systems, or df
   and "vos partinfo", the problem was in the way the vice partition
   was created.

   You want to check the default size of the 

	bytes per inode
	
   Please check this setting on your fileserver's vice partitions  to 
   make sure they are in sync.

   We had a site which had a volume which contained 800,000 files
   which all had a size around 100 bytes.  And they had a default
   "bytes per inode" at 4096.  So this resulted in about 4K of wasted
   space per file and it took up all the space.

   So this is a good candidate to check.

 - And you can also remove the readonly on the RW site and then do
   the addsite and release again to see if this helps.

Thanks

Todd