Route Table scenario (Backend Cluster):
The Route Table protection option in the Recovery Kit for EC2 provides the ability to automatically update the routing in the VPC. During a failover the recovery kit will update the route table to reflect the new Elastic Network Interface (ENI) location of the virtual IP address on the target server. In order for LifeKeeper to protect, monitor and update the Route Table in the VPC, the following configuration steps must be performed (this also applies to the active/active configuration):
- The virtual IP address to be protected by the LifeKeeper for Linux IP Recovery Kit must be out of range of the allocated CIDR in the VPC.
- The virtual IP address must be protected by LifeKeeper prior to creating the Recovery Kit for EC2 resource.
- The Source/Dest Checking of the ENI must be disabled. This is required in order for the instance to accept network packets for the virtual IP address.
- Broadcast PING checking of the LifeKeeper IP resources must be disabled. LifeKeeper monitors IP resources by executing the Broadcast PING test of the IP address on the local subnet. In multiple availability zone environments this feature would not be useable because of the different subnets that exist between multiple availability zones. To disable this feature you must set the NOBCASTPING entry in the /etc/default/LifeKeeper configuration file as follows:
- The Route Table should have a route entry for the virtual IP address and the ENI of the active server.
Note: Since the EC2 recovery kit will protect the configuration of the Route Table once the corresponding EC2 resource has been created, the Route Table should not be modified manually after hierarchy creation.
Destination: VIP 10.1.0.10/32
Elastic IP scenario (Frontend Cluster):
The Elastic IP (EIP) protection option in the Recovery Kit for EC2 provides the ability to automatically re-associate an EIP with a specific ENI (the ENI used by EC2 resource on the active or backup server).
In order for LifeKeeper to protect, monitor and update the association of an EIP with the ENI on the active or backup server, the following configuration steps must be performed:
- One ENI can be associated with only one Elastic IP. No other EIPs (any EIPs other than the one used by EC2 resource) should be associated with the specific ENIs. Otherwise the recovery kit will disassociate any other EIPs that are already associated with the specific ENIs.
- Since an Elastic Block Store (EBS) of AWS can only be attached to one EC2 instance, DataKeeper for Linux is recommended when creating an HA cluster configuration using EBS.
- We recommend increasing RESRVRECTIMEOUT in /etc/default/LifeKeeper to 300 from 150 as the default. RESRVRECTIMEOUT is the number of seconds that a LifeKeeper process will sleep when waiting to reserve a resource for “recovery”, while another process already has the resource reserved.