part 11 · Cluster & Operations

36 · Recovery & Replication

Failure is part of the architecture. Detect, find a valid copy, recover, verify checksums, mark healthy.

What happens after the crash? Answered by Translog replay and copy-based recovery.

In this lab: Crash the engine at different points and recover it, then corrupt a file and watch verification fail.

loading mind map…
building the index…