|
I am having problens configuring this kind of setup. Heartbeat works fine,
but i guess i am getting wrong in the logic.
What i want!
2 load-balancer with HB and ldiretord
2 Postfix.
Think about this.
have 2 shared ips for the load-balancer.
one external and one internal.
will call lb1 and lb2 for load-balancers and pf1 pf2 for postfix
ips are just illustratives
lb1:
shared external ip 222.222.222.222 in eth0
shared internal ip 192.168.0.1 in eth2 ( this will be the gateway for pf1
e pf2 )
internal ip 192.168.0.3
lb2:
shared external ip 222.222.222.222 in eth0
shared internal ip 192.168.0.1 in eth2 ( this will be the gateway for pf1
e pf2 )
internal ip 192.168.0.2
pf1:
internal ip 192.168.0.11
gateway 192.168.0.1
pf2:
internal ip 192.168.0.12
gateway 192.168.0.1
Ldirectord checks the pf and etc. This step works fine.
Now the problem is the heartbeat.
I am doing ucast. over interface eth2 in both but look the problem.
The heartbeat is checking ok if the lb1 shutdowns.
But problens begins when:
lb1 loses eth0, heartbeat simple does nothing.
lb1 loses eth2, heartbeat in lb2 think lb1 had going down, and takeover, but
now we have conflict ips.
How to solve this ?
Now researching more, i find out that heartbeat can do check in 2
interfaces. But problem still hapens, just will be harder to happen.
I guess if i introduce more one node, it may do the tricky. Cause now we
have 3 nodes, and they can never have a tie. and becamo with same ip. But
this new heartbeat will only have heartbet without ldiretord.
I am current not using stonith, should i use ?
If anyone have any tips please i would be glad to hear them!!
Thanks!
[]'sf.rique
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users
|
