--- vrrpd-1.0.orig/README
+++ vrrpd-1.0/README
@@ -0,0 +1,70 @@
+vrrpd version 1.0 2002/02/05
+============================
+
+The Monitored Interface extensions to vrrpd were originally
+implemented to provide support for failover between a pair
+of state-synchronised Checkpoint Firewall-1 modules. These
+servers were HP LT6000r machines with two quad ethernet
+NICs each and Red Hat Linux 7.0 (2.2 kernel required by
+Firewall-1 v4.1 SP5).
+
+Full credit to David Moore of Hewlett-Packard (NZ) Ltd
+for initiating the project that made this necessary.
+
+Anyway, an example command line for a primary server is;
+
+  vrrpd -i eth0 -v 10 -p 200 -m "eth1 eth2 eth3" -c 100
+
+This specifies eth0 as the interface to run the VRRP process
+on, a VRID of 10, a priority of 200, that interfaces eth1, eth2
+and eth3 are to be monitored for failure, and that if one of
+these interfaces does fail, the priority is to be reduced by 100.
+
+This allows the vrrpd process below (on the standby) to win;
+
+  vrrpd -i eth0 -v 10 -p 150
+
+as the master will advertise a priority of 100 while any of eth1-3
+are disconnected or administratively down. Note that there is
+little point in monitoring interfaces on the secondary UNLESS
+you are load balancing PAIRS of interfaces across different
+servers.
+
+This configuration has been tested as providing transparent
+failover for HP-UX systems as well as routers, with up to five
+vrrpd daemons running concurrently on separate interfaces.
+Each vrrpd process was monitoring the other four interfaces,
+and was using the multicast VRRP MAC address so that arp
+caching was not an issue. Intel 82559-based cards (eepro100
+driver) were used.
+
+During a failure (ie: ifconfig ethX down, pull cable out, turn
+switch off, etc.) there is a 3x advertisement delay while the
+VRRP election takes place. At the default advertisment interval
+(1 sec), this was not long enough to cause TCP streams to break.
+
+Interfaces are monitored in two different ways. The IFF_UP flag
+is checked to detect an administrative shutdown, and the link
+status bits of the MII transceiver status word are checked to
+detect a failed cable, switch or port. This should work with
+most Fast Ethernet drivers, but not with 10Mbps-only cards.
+Gigabit NICs have not been tested.
+
+There is no way to have different priority deltas for the
+different monitored interfaces at present. However, it would
+not be difficult to implement.
+
+As a last point, if you are using this on a firewall (eg: FW-1),
+do not forget to allow VRRP and IGMP traffic from the 'original'
+IP addresses of all interfaces running vrrpd to destination
+address 224.0.0.18, otherwise elections will not work.
+
+Comments are welcome, I guess...
+
+
+David Hunter
+Technical Consultant
+gen-i limited
+Auckland, NZ
+<david.hunter@gen-i.co.nz>
+