[OpenAFS] Two questions AFS as a *large* webserver system backend in lieu of NFS

[Michael Loftis]

The interop isn't an issue for us, all the Linux boxes are the same already 
as necessity anyway.

We already have the problem of having to mount the jail inside of the udir, 
which is a huge problem for us already (Linux isn't handling 2,500+ mounts 
all that greatly well).

Backwards compatibility isn't really an issue for us either since this 
system will be private.

I appreciate very informative comments I've gotten so far, keep them 
coming! :)

--On Friday, July 25, 2003 13:40 -0500 David Thompson <thomas@cs.wisc.edu> 
wrote:

>
> In cases where we have needed devices in /afs, we mount a tmpfs file
> system  over /dev in the jail (in afs), and populate it as needed.
>
> This may or may not work for you.  As was pointed out, devices in /afs
> would  have their own set of problems...
>
> Dave
>
>
> Michael Loftis wrote:
>> Someone mentioned privately AFS can't/won't handle device nodes?  Is
>> this  (still) true?  If so, why, and how difficult would it be to
>> develop this  feature?  Without it almost all the utility of AFS is lost
>> to me.  I need  to have user jails, with a 'shared' part that doesn't
>> change (you know  /usr/bin/usr/sbin) replicated to everyone without
>> having to keep 300MB per  user (actually almost 3x that I think).
>
>



--
Michael Loftis
Modwest Sr. Systems Administrator
Powerful, Affordable Web Hosting