Child pages
  • Single point of failure
Skip to end of metadata
Go to start of metadata

Tests that the varios mechanisms for avoiding single point of failures works correctly.
Note that the test found here influences the entire system, and should therefore not be run together with other tests.

Message bus failover test

Prerequisites: The system is running with 2 networked networked brokers and the client has been configured to use failover between the brokers.

  1. Start a operation, and stop the primary broker. Confirm that the client and pillars failsover to the secondary broker and finishes the operation here.
  2. Start the primary message bus up again.
  3. Start a operation, and stop the secondary broker. Confirm that the client and pillars failsover back to the primary broker and finishes the operation here. 
Missing pillar 
  1. Shutdown one of the pillars.
  2. Configure the client to have a reasonable short time out for identifies and conversations (1 minute),
  3.  Run through all the steps described in the Sanity test and confirm the system still works correctly (but somewhat slower, as every operation has to wait for the operation timeouts).
  • No labels