Principle:Apache Druid Supervisor Operational Management
| Knowledge Sources | |
|---|---|
| Domains | Streaming_Ingestion, Operations |
| Last Updated | 2026-02-10 00:00 GMT |
Overview
A supervisor lifecycle management principle that provides operational controls for suspending, resuming, resetting, terminating, and adjusting consumer offsets of streaming supervisors.
Description
Supervisor Operational Management provides the day-to-day operational controls for managing streaming supervisors:
- Resume: Restart a suspended supervisor (POST /suspend → POST /resume)
- Suspend: Pause ingestion without losing state or offsets
- Hard Reset: Clear all supervisor state and restart from scratch
- Terminate: Permanently stop the supervisor and clean up resources
- Set Offsets: Manually adjust consumer offsets/sequence numbers for specific partitions
- Handoff Early: Force current tasks to publish segments immediately
Bulk operations (resumeAll, suspendAll, terminateAll) are also available for fleet-wide management.
Usage
Use this principle for operational management of running streaming supervisors. Actions are accessible from the supervisor row context menu or action dialog in the Supervisors view.
Theoretical Basis
Supervisor lifecycle management follows a state machine with actions pattern:
Supervisor state machine:
RUNNING ↔ SUSPENDED (via suspend/resume)
RUNNING → TERMINATED (via terminate, irreversible)
ANY → RESET (via reset, clears all state)
Actions:
POST /druid/indexer/v1/supervisor/{id}/resume
POST /druid/indexer/v1/supervisor/{id}/suspend
POST /druid/indexer/v1/supervisor/{id}/reset
POST /druid/indexer/v1/supervisor/{id}/terminate
POST /druid/indexer/v1/supervisor/{id}/resetOffsets
{ type: 'end', partitions: { type: 'end', stream, partitionOffsetMap } }
Bulk:
POST /druid/indexer/v1/supervisor/resumeAll
POST /druid/indexer/v1/supervisor/suspendAll
POST /druid/indexer/v1/supervisor/terminateAll