[OpenAFS] key issues for OpenAFS future?

[Warren.Yenson@morganstanley.com]

> > OpenAFS works fine with journalled filesystems. The only incompatability
> > is that you cannot use reiserfs as your cache partition with 2.4.x+
> > kernels since reiserfs does not support retrieving a file from disk from
> > just the inode number, unlike most other unix filesystems.
> >
> > Ext3 works just fine as a cache partition, and ANY of the filesystems
> > works for server vice partitions.
>
> As far i know you cant turn on logging on your /vice - partitions on
> solaris if dont want things to break. You say it works now ? Do i have to
> run namei-filerserver then ?

It's not an apples to apples comparison.  There is no native-UFS
fileserver implementation for Linux, only a NAMEI implementation.  As
such, it does not matter which filesystem you use for a /vice filesystems
on Linux fileservers.

On Solaris fileservers, the default is UFS /vice partitions, and no, you
cannot then turn on UFS logging (unless you like wholesale data
corruption) with these servers.  If you decide to use the NAMEI fileserver
on Solaris, you can use whichever filesystem you like.  Note that we have
observed some penalty in performance using a NAMEI fileserver as opposed
to the UFS fileserver.