Advanced level · Module 4: Replication, Availability and Recovery · Lesson 2 of 4
High Availability and Failover
Plan how a standby takes over when the primary fails, and avoid split-brain.
In this lesson you will
- Perform a planned failover
- Explain automatic failover and split-brain risks
Key concepts
promotionplanned vs unplanned failoversplit-brainfailover tools
The lesson
Failover promotes a standby to become the new primary. In a planned failover — for maintenance — you stop writes, let the standby catch up and promote it.
Automatic failover tools detect a failed primary and promote a standby, but must avoid split-brain: two servers both accepting writes, creating conflicting data. That is why they use quorum and fencing.
Applications need to find the new primary, through a changed address, a proxy or connection settings listing several hosts. Practise failover regularly so it is routine.
Step by step
- Plan a maintenance failover step by step.
- Promote the standby.
- Repoint the application to the new primary.
- Rebuild the old primary as a standby.
Practice activity
Perform and document a planned failover of the health database.
Expected result: A completed failover with the steps, timings and any problems recorded.
Check your understanding
-
What is split-brain?
Show answer
Two servers both acting as primary and accepting writes.
-
What must happen to applications after failover?
Show answer
They must connect to the new primary.
Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.