KB21584 - Cluster VIP flapping between both of the nodes in Active/Passive cluster

This article provides information about an issue of cluster VIP flapping between nodes in Active/Passive cluster, possible cause and resolutions.

When two nodes are in a Active / Passive cluster, this may result in unexpected changing of VIP ownership between the two devices.  This will impact end users as their user sessions will need to reconnect to the new VIP owner.

 

In an Active/Passive configuration, the PCS uses a Virtual IP (VIP) that is owned by only one node (considered the active node) at any given time. The other node in the cluster would be the passive node. Each node in a cluster is in near-constant communication with the other to ensure each node is reachable.  This task is performed by sending Address Resolution Protocol (ARP) pings and confirming a successful response is received.  

In the event that a node does fails (timeout being the gateway's ARPing timeout + 5 seconds), the passive node will assume ownership of the VIP and sends a gratuitous ARP message notifying other networking devices that it now owns the VIP address.


When the VIP is flapping between two nodes, this can occur due to the following reasons:

• If the default gateway cannot pass ARP (Keep Alive Pings) pings between the two nodes, to check if both of the nodes are alive or not.
• If both of the nodes in the cluster cannot reach the default gateway
• Issues with duplicate IP addresses assuming ownership of the VIP.  For more details, refer to KB21297 - Both members of an Active-Passive cluster report that the partner PCS has the VIP address  

To resolve this issue, ensure both nodes can reach the default gateway.