Product Documentation

High Availability

Sep 30, 2013
What are the various ports used to exchange the HA-related information between the nodes in an HA configuration?
In an HA configuration, both nodes use the following ports to exchange HArelated information:
  • UDP Port 3003, to exchange heartbeat packets
  • Port 3010, for synchronization and command propagation
What configurations are not synced or propagated in an HA configuration in either INC or non-INC mode?
Configurations implemented with the following commands are neither propagated nor synced to the secondary node:
  • All node specific HA configuration commands. For example, add ha node, set ha node, and bind ha node.
  • All Interface related configuration commands. For example, set interface and unset interface.
  • All channel related configuration commands. For example, add channel, set channel, and bind channel.
Note: For more information about HA Configuration in INC mode, see Configuring High Availability Nodes in Different Subnets.
What configurations are not synced or propagated in an HA configuration in INC mode?
The following configurations are neither synced nor propagated. Each node has its own.
  • MIPs
  • SNIPs
  • VLANs
  • Routes (except LLB routes)
  • Route monitors
  • RNAT rules (except any RNAT rule with VIP as the NAT IP)
  • Dynamic routing configurations.
What are the conditions that trigger synchronization?
Synchronization is triggered by any of the following conditions:
  • The incarnation number of the primary node, received by the secondary, does not match that of the secondary node.
    Note: Both nodes in an HA configuration maintain a counter called incarnation number, which counts the number of configurations in the node's configuration file. Each node sends its incarnation number to each other node in the heartbeat messages. The incarnation number is not incremented for the following commands:
    • All HA configuration related commands. For example, add ha node, set ha node, and bind ha node.
    • All Interface related commands. For example, set interface and unset interface.
    • All channel-related commands. For example, add channel, set channel, and bind channel.
  • The secondary node comes up after a restart.
  • The primary node becomes secondary after a failover.
Does a configuration added to the secondary node get synchronized on the primary?
No, a configuration added to the secondary node is not synchronized to the primary.
What could be the reason for both nodes claiming to be the primary in an HA configuration?
The most likely reason is that the primary and secondary nodes are both healthy but the secondary does not receive the heartbeat packets from the primary. The problem could be with the network between the nodes.
Does an HA configuration run into any issues if you deploy the two nodes with different system clock settings?
Different system-clock settings on the two nodes can cause the following issues:
  • The time stamps in the log file entries do not match. This situation makes it difficult to analyze the log entries for any issues.
  • After a failover, you might have problems with any type of cookie based persistence for load balancing. A significant difference between the times can cause a cookie to expire sooner than expected, resulting in termination of the persistence session.
  • Similar considerations apply to any time related decisions on the nodes.
What are the conditions for failure of the force HA sync command?
Forced synchronization fails in any of the following circumstances:
  • You force synchronization when synchronization is already in progress.
  • The secondary node is disabled.
  • HA synchronization is disabled on the current secondary node.
  • HA propagation is disabled on the current primary node and you force synchronization from the primary.
What are the conditions for failure of the sync HA files command?
Synchronizing configuration files fail if the secondary node is disabled.
In an HA configuration, if the secondary node takes over as the primary, does it switch back to secondary status if the original primary comes back online?
No. After the secondary node takes over as the primary, it remains as primary even if the original primary node comes back online again. To interchange the primary and secondary status of the nodes, run the force failover command.
What are the conditions for failure of the force failover command?
A forced failover fails in any of the following circumstances:
  • The secondary node is disabled.
  • The secondary node is configured to remain secondary.
  • The primary node is configured to remain primary.
  • The state of the peer node is unknown.