Health monitor

Health monitoring is the act of checking a set of pool members for "aliveness". This usually means HTTP, TCP, UDP, or ICMP connectivity is possible to the node.

In MidoNet’s case, only TCP connectivity is checked. Health monitors work by sending packets to the pool members and checking whether or not they receive a reply. The node is considered ACTIVE if the pool member responds to the packet within a certain amount of time, and after a certain amount of retries. Therefore, health monitors act on the following three variables:

  • max_retries: How many times the health monitor sends a packet to the pool member without receiving a response before the health monitor considers the node to be INACTIVE
  • delay: Amount of time between each transmission of a packet from the health monitor to the pool member
  • timeout: Additional timeout after a connection has been established

The health monitor keeps track of the current state of all pool members it is assigned to. Load balancing decisions can then be made based on the "aliveness" of a pool member.

Questions? Discuss on Mailing Lists or Chat.
Found an error? Report a bug.

loading table of contents...