director failover with problems
clemens.draschl at techconsult.at
Mon Aug 27 10:29:23 MDT 2001
maybe a stupid question, but i read already the faq, the howto, the
the following setup exists:
2 director nodes
working with heartbeat 0.4.9, linux 2.2.17 and the lvs with ipvsadm
director one: external ip 172.16.2.161, internal 192.168.10.1
director two: external ip 172.16.2.162, internal 192.168.10.2
director cluster ip: 172.16.2.160
using nat for directors
the heartbeat goes over two serial lines, working fine.
the problem is, after killing the primary director, the backup director
takes over the virtual cluster ip, but not the internal. due to the
default gw (192.168.10.1, director1 internal) set up on each webserver,
each reply goes through director1, although it doesn't exist any longer
(e.g. power kill).
can anybody help me?
ps: sorry, if someone already had this problem, but i'm on this lost for
only one day. so please forgive me. :-)
More information about the Linux-HA