On Wed, Mar 17, 2004 at 04:25:06PM +0100, Andreas Schuldei wrote: > * Ragnar Wisloff ([EMAIL PROTECTED]) [040317 10:34]: > > There's been an entertaining discussion about this :-) > > Let me just make the comment that too many groups will create real > > problems, no matter what the naming scheme is. Since NFS will honour no > > more than 16 groups, users logging in to a thin client server (which > > mounts the home dirs etc. using NFS) will risk not being able to use > > common file storage set aside for a specific group. We have diagnosed > > this problem at one school at least, where one of the teachers was a > > member of 28 groups. > > yes, that is a protocol problem. it is still there in nfs v3. > perhaps it is lifted in v4, but we would not want to use that > since it is still bleeding edge. > there were patches since 1999 to change this but they would have > broken the protocoll. i suggest we use them. > just kidding. btw: smbfs does not suffer from *that* restiction. > ironically enough, the groups that i introduced can *help* to > avoid that limit, since they dont necessarily need to be > posix-groups. we can use posix groups if file-sharing is > necessary and others to do organisational and structural > partitioning.
Then I would opt for changing the groups juadmin and genagegp so that they no longer is added by default. Also to cause less confusion to the ones adding users and groups, I vote for making this groups optional in wlus. When you add a user, you may create this additional age groups, and groups like juadmin, but you don't need to. -- Finn-Arne Johansen [EMAIL PROTECTED] http://bzz.no/
