Skip to main content
PreparationZone Logo
PrepZone

Rolling Upgrades and KRaft Migration

Upgrade Kafka 3.7 → 4.x with zero downtime — rolling broker restarts and KRaft migration checklist.

Pass checkpoint below

Why this matters

  • VaultCommerce upgraded 3.6 → 3.7 KRaft with zero client downtime.
  • Upgrade order: controllers one at a time, then brokers; verify quorum between each.
  • Inter-broker protocol and log message format version bump after all nodes updated.
  • Kafka 4.x removes ZK — migration checklist is mandatory reading.
Producer
Leader P0Broker 1
Follower P0Broker 2
Follower P0Broker 3
Each partition has one leader and N-1 followers. Only the ISR set can become leader on failover.

Rolling restart procedure

  1. Verify cluster healthy (no URP). 2) Upgrade binary on broker N. 3) Restart broker N. 4) Wait ISR sync. 5) Repeat. VaultCommerce maintains runbook with 15-minute soak between nodes. Never restart >1 controller simultaneously.

Key points

  • Rolling restart — one broker at a time to preserve quorum
  • inter.broker.protocol.version — wire protocol between brokers
  • log.message.format.version — on-disk message format
  • KRaft migration — ZK to KRaft one-way path with dual-write phase
  • Soak period — validation window before next node restart

The rest of this guide is part of Pro

Free covers every fresher guide in full. Pro unlocks Senior, Staff and Architect guides across every track.

Still to come · 2 sections

  • Post-upgrade version bump
  • VaultCommerce rollout checklist

Test yourself

Answer these before moving on — recall is what makes it stick.