On Thu, 2004-12-02 at 08:21 -0500, Paul W. Frields wrote: > On Wed, 2004-12-01 at 19:25 -0500, Charles R. Anderson wrote: > > On Wed, Dec 01, 2004 at 05:42:50PM -0500, Paul W. Frields wrote: > > > On Wed, 2004-12-01 at 14:02 -0800, Florin Andrei wrote: > > > > Perhaps don't leave the *PORT variables empty, but add sensible values? > > > > > > Sensible values will be widely variant from one location to the next. A > > > "sensible value" is one that gets through *your* firewall. What's there > > > should be good enough for even the laziest admin, IMHO, and yes, I > > > resemble that remark too. :-) > > > > There are "standard" port numbers for some of these: > > > > MOUNTD_PORT=635 > > NFSD_PORT=2049 > > For rpc.nfsd, you're right on. Sorry about that. For rpc.mountd and > rpc.statd, the man pages show that ports are assigned randomly by > portmap. Is that not correct? As a correction to my correction, the current initscript for nfs doesn't make any allowance for changing the listening port for nfsd. This is not a value judgment, just an observation. Bill Nottingham: Would you agree that bugs for these config files should go, then, against the packages to which they refer (e.g. nfs-utils for /etc/sysconfig/nfs)? Do you have any insight into whether the maintainers would be interested in even doing this? I'll be happy to file the tracker bug and the candidate for /etc/sysconfig/nfs to get things started, if so. -- Paul W. Frields, RHCE
