andrew at beekhof
Aug 7, 2012, 4:13 AM
Post #2 of 2
(152 views)
Permalink
|
On Tue, Aug 7, 2012 at 1:42 AM, Andy Furtado <awf7d0 [at] yahoo> wrote:
> Hello,
>
>
> Is it possible to setup an n-to-1 cluster configuration and have heartbeat manage a different VIP for each virtual pair.
> The n-to-1 configuration would have a single slave node, able to take over for any one of the failed N masters at a time.
You'll want pacemaker on top of heartbeat for that.
http://www.clusterlabs.org
> In this configuration each masternode would have a staticip-addr and a VIP. When the master fails, the VIP for that master is configured on the slavenode, and the slave acts as that master.
> Once the slavenode is acting as a master, it remains in this state and cannot takeover of another failed master until the original masternode is restored, and the original slave transitions back to the slave state.
>
>
> Example configuration
> masternodeA
> staticip:10.1.1.1
> VIP:10.1.1.101
>
>
> masternodeB
> staticip:10.1.1.2
>
> VIP: 10.1.1.102
>
>
> slavenode
> staticip:10.1.1.3
>
>
> If masternodeA fails, slavenode becomes active as masternodeA and is configured with VIP 10.1.1.101
> If masternodeB fails, there is no failover available since slavenode is currently acting as masternodeA
>
>
> When masternodeA is restored, slavenode releases VIP 10.1.1.101, and is now ready to take over for either masternodeA or masternodeB.
>
>
> I understand this is not an ideal fail over solution, but one I must live with until further design can be done.
>
> I've searched the internet, and the HA mailing lists without much success.
>
> Any info, or input would be appreciated.
>
>
> Best Regards,
> Andy
> _______________________________________________
> Linux-HA mailing list
> Linux-HA [at] lists
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
Linux-HA [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
|
