[700] in linux-net channel archive
Netgroup-aware 'rwall'. (ALPHA)
daemon@ATHENA.MIT.EDU (Jeff Uphoff)
Sun Jul 16 11:35:33 1995
To: submit-linux-dev-net@ratatosk.yggdrasil.com
From: juphoff@tarsier.cv.nrao.edu (Jeff Uphoff)
Followup-To: linux.dev.net
Date: 16 Jul 1995 08:06:12 GMT
I've written an NIS-netgroup-aware 'rwall'. It's intended primarily for
Linux, but it's tested out OK on Solaris 2.4 under both GCC and Sun's
bundled compiler. Other OS's are untried as yet, though SunOS 4.1.x is
known to still be a problem. (I've not worried about portability much
yet...)
Primary advantages over the standard Linux (and Sun in some cases)
'rwall' program:
1) Aware of NIS netgroups.
2) You can specify multiple hosts and/or netgroups to broadcast to, with
any combination of the two (and as many of each) that your heart
desires.
3) Eliminates duplicate host broadcasts. If you have a host that
appears in multiple netgroups (and you send to more than one--thus
"including" that host more than once), or that appears in a single NIS
netgroup more than once (say by existing in both fully-qualified form
and in "short" form), it will only receive the 'rwall' message once; I
keep a list of "done" hosts and thus eliminate duplicates. *This
feature can be disabled by specifying "-d" on the command line.*
4) Command-line-configurable timeout. One thing that annoys me about
Sun's (and Linux's) 'rwall' is that, the way it does its RPC calls, it
waits for the system-defined RPC timeout (long!) on each "unresponsive"
host before going on to the next. (In large netgroups, with many
"phantom" hosts, this can be *extremely* annoying.) I've used a
different RPC call that allows better control of timeout(s), and a
default ("local") timeout value is both compiled into the base code and
then is over-rideable on the command-line. The default in the
distributed code is 5 seconds. (I'm impatient by nature.)
This code is ALPHA, and was just whipped out in the past couple of days.
As with the current 'rwall', either STDIN or a file can be used as a
source for the message to broadcast. The message-handling code is
pretty much a crib from the current (BSD-based) Linux 'rwall' (with some
Berkeleyisms removed to pacify Solaris compilers)--this will be removed
in the next version as I finish from-scratch coding and make this GPL'd.
All of the RPC, duplicate-tracking, etc. code is original by me.
Anyway, bug reports much appreciated. Things work swimmingly for me,
but that never means much when you're the author. :)~
Available at:
linux.nrao.edu:/pub/people/juphoff/nrwall/
--Up.
P.S. If someone else has written a netgroup-aware 'rwall' for Linux I'd
love to trade notes with them. I don't know of any out there yet--so
that's why I wrote this one.
--
Jeff Uphoff - systems/network admin. | juphoff@nrao.edu
National Radio Astronomy Observatory | jeff.uphoff@linux.org
Charlottesville, VA, USA | http://linux.nrao.edu/~juphoff/
