Backup and Restore Runbook

This page focuses on practical backup and restore steps.

Use it before risky changes: upgrades, token rotation, transport changes, host migration, or switching between Bedrock and Java-side topologies.

What to back up

Minimum:

  • config/ServerProperties.json
  • service wrapper files
  • deployment notes for ports and tokens

Recommended:

  • previous release artifacts
  • logs around the last good known state
  • plugin configs such as GeyserVoice/config.yml
  • Bedrock world pack config files
  • addon package versions currently installed in the world
  • service manager files such as systemd units or panel startup commands
  • runtime override notes if startup flags are used

Before changing anything

Take a snapshot when you are about to:

  • upgrade VoiceCraft
  • change transports
  • rotate tokens
  • switch topology
  • change host bindings or firewall rules
  • move from plugin-managed runtime to external runtime

Restore workflow

  1. Stop the affected service.
  2. Restore ServerProperties.json.
  3. Restore related plugin or addon config if the topology changed.
  4. Restore the matching addon/plugin package if version compatibility matters.
  5. Restart VoiceCraft.
  6. Restart or reload the Minecraft-side integration.
  7. Validate transport auth and bind flow.

What a restore does not fix automatically

  • firewall mistakes
  • DNS or host reachability issues
  • mismatched client or plugin config
  • topology mistakes after a network redesign
  • a provider that blocks the required transport path
  • players using a newer incompatible client package

Validation after restore

Check:

  1. server starts cleanly
  2. chosen transport is enabled
  3. token matches the integrating node
  4. player bind and audio flow work again
  5. server commands show expected clients/entities
  6. logs no longer show the error that triggered the restore

Backup naming

Use names that include:

  • date
  • VoiceCraft version
  • topology
  • reason

Example:

2026-05-13-voicecraft-1.6.1-bds-before-token-rotation

Good names matter during incidents because they make it obvious which backup belongs to which topology.