Scenarios

Task-oriented playbooks for using CritterWatch to handle real production situations. Each scenario starts from a symptom ("a downstream system is broken", "messages are piling up", "a projection is behind") and walks through diagnosing and remediating it with CritterWatch — the dashboard, the MCP tools, or the CLI.

These are how-to guides. For reference material on individual screens see the UI Reference, and for the operator command catalog see Inbound Commands.

Available scenarios

• A downstream system is broken — pause and resume a listener — stop consuming from a queue while a dependency is down, then resume cleanly once it recovers.

Suggested scenarios (backlog)

These are planned but not yet written — contributions welcome:

• Messages are piling up on a queue (back pressure vs. a stuck listener)
• A projection has fallen behind or is stuck — pause, rebuild, or rewind
• A bad deploy is poisoning a queue — drain, hotfix, replay the dead letters
• A node won't leave the cluster cleanly (eject a dormant node)
• Right-sizing alert thresholds from observed baselines