|
Hi to all,
First of, many thx for the replies...
To Joseph Mack NA3T:
I'm actually implementing high-availability for load balanced
firewalls. If one firewall node goes down, then a standby node takes
up the job. That's what i'm basically trying to implement.
To Graeme:
You guessed right..I am using load balancing. What I really want to do
is that the 2 standby nodes take over the active nodes in a sequence,
i.e, if any of the 3 active nodes fails, then one standby node (say
BACKUP 1) takes over; if another of the active nodes fails (including
the newly standby node that has become master), then the second
standby node (say BACKUP 2) takes over. I dont want the active nodes
to get a second VIP on them but rather have a standby node take it.
Consider the following:
MASTER A: VIP1 priority 100
MASTER B: VIP2 priority 100
MASTER C: VIP3 priority 100
BACKUP 1: VIP1 priority 75
VIP2 priority 50
VIP3 priority 50
BACKUP 2: VIP1 priority 50
VIP2 priority 75
VIP3 priority 75
In normal operation, VIP1 is on MASTER A, VIP2 MASTER B, VIP3 MASTER C.
If node1 fails (alone), BACKUP 1 takes VIP1.
If node2 fails (alone), BACKUP 2 takes VIP2.
If node3 fails (alone), BACKUP 2 takes VIP3.
If the event that 2 nodes fail, then there is a problem in one out of
the 3 possible scenarios. This is the actual problem I'm talking
problem with. When MASTER B & C fail simultaneously, both VIP2 & 3 go
to BACKUP 2. Is there a way to prevent this from happening? Are the
notify scripts of any help? Can we change the priority values
dynamically (i.e. while keepalived is running)?
Warm regards,
Visham
|
