Failover servers i.e Heartbeat

[gordon]

Hi,

We use Heartbeat on our asterisk servers so that if asterisk server 1 fails asterisk server 2 kicks into life.

This works by basically
Asterisk 1 has ip 10.0.0.4 as its base ip and 10.0.0.10 on the same nic
Asterisk 2 has ip 10.0.0.6 as its base and if it cannot see 10.0.0.4 it adds 10.0.0.10 to its nic, when it can see 10.0.0.4 again it drops the 10.0.0.10 off again.

Is there away to do that with Zeroshell, however i guess its going to be a lot more complicated as i have 60 vlans on the zeroshell box with pppoe connections.

The idea is so that if Zeroshell 1 dies Zeroshell 2 automatically takes it’s place.

Just a thought guessing the answer is gonna be a no not really!

Cheers
G

[atheling]

Interesting concept. Basically a subset of clustering which is not easy to make robust in the general case….

There is a hook for a “pre boot” script to be run. You might be able to put something in there that expects heartbeats from the master. If it sees them then it keeps waiting. When heartbeats fail the the script terminates allowing the “boot” process to occur.

On the master side, spawning a daemon in the post boot script could generate heartbeats.

Lots of details to nail down to make it robust. But it might be possible.

[gordon]

Cheers,

certainly a thing or two to experiment with.

Will have a play when i get time and let you know.

G

[Author]

Posts