Occasionally a '#cvs.lock' directory gets rsync'd, which then
causes a problem, when I try to cvs update against my rsync'd
repository.  Rather than every person who uses rsync having to
add "--exclude '#cvs.lock'" to their rsync options, would it make
sense to have this done on the rsync server?

Also, how are other people using rsync?  I currently just rynsc
into /u0/netbsd/, and then on an internal client I set my CVSROOT
to :pserver:root@my.rsync.box:/u0/netbsd/anoncvs/main.
After the initial 'cvs checkout src', I just run 'cvs update -d -P'
in /usr/src on the client.  Unfortunately I also need to do a 
make cleandir or things like 'gdb' will cause cvs to fail. (e.g.
it wants to make a gnu/usr.bin/gdb/gdb directory, then prune it?)

So the whole process looks like:
server#	rsync -avz --delete --exclude '#cvs.lock' rsync://sup.netbsd.org/anoncvs anoncvs
client# cd /usr/src
client# make cleandir
client# cvs update -d -P

If I don't do the 'cleandir' first, cvs will fail to update, unless I omit
the '-d' option.  If I omit the '-d' option, I don't get any new
subdirectories, and a subsequent 'make build' will/may fail.  I would
prefer to not 'cleandir' first, so I can easily update without a full
build cycle.

This seems like it should be easy, but I'm not a cvs wizard, and so
far I've used the above process.

-Andrew
-- 
-----------------------------------------------------------------
Andrew Gillham                            | This space left blank
gillham@whirlpool.com                     | inadvertently.
I speak for myself, not for my employer.  | Contact the publisher.