apache/druid · error · IllegalArgumentException

Reset with skipped offsets is not supported when…

Error message

Reset with skipped offsets is not supported when useEarliestOffset is true.

What it means

SupervisorManager.validateResetAndBackfill rejects a reset-and-backfill request when the seekable stream supervisor is configured with useEarliestOffset=true. In that mode the supervisor already starts from the earliest available offsets, so an explicit reset that skips offsets is contradictory and unsupported. The check runs before any offsets are reset, so state is left unchanged.

Solutions

  1. Set ioConfig.useEarliestOffset=false (suspend supervisor, update spec via POST /supervisor, resume) and retry the reset-and-backfill
  2. Use the plain reset endpoint (which resets to earliest/latest per current config) instead of the skipped-offsets backfill flow
  3. If earliest-data ingestion is what you want, drop the backfill request — data is already being read from earliest offsets

Example fix

// before
{"ioConfig": {"topic": "wiki", "useEarliestOffset": true}}
// after
{"ioConfig": {"topic": "wiki", "useEarliestOffset": false}}
// then: POST /druid/indexer/v1/supervisor/<id>/reset
Defensive patterns

Strategy: validation

Validate before calling

const spec = await (await fetch(`${overlord}/druid/indexer/v1/supervisor/${id}`)).json();
if (spec.spec.ioConfig.useEarliestOffset) throw new Error('reset-with-backfill unsupported while useEarliestOffset=true');

Type guard

function supportsResetBackfill(spec) { return spec?.spec?.ioConfig?.useEarliestOffset === false; }

Try / catch

try { await resetBackfill(id); } catch (e) { if (/useEarliestOffset is true/.test(e.message)) { updateSpecDisableEarliestOffset(id); } else throw e; }

Prevention

When it happens

Trigger: POST /druid/indexer/v1/supervisor/<id>/reset (with resetOffsetsAndBackfill-style API, resetToLatestAndBackfill path) on a supervisor whose ioConfig.useEarliestOffset is true.

Common situations: Operator created the supervisor with useEarliestOffset=true and later tries a reset-and-backfill to a specific offset; copy-pasted spec template with useEarliestOffset set; migration from an earlier behavior where the combination was tolerated.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/f4065232c69cb11a. Report an issue: GitHub.

Appendix: source

Thrown at indexing-service/src/main/java/org/apache/druid/indexing/overlord/supervisor/SupervisorManager.java:512

    SupervisorTaskAutoScaler autoscaler = autoscalers.get(id);
    if (autoscaler != null) {
      autoscaler.reset();
    }

    return ImmutableMap.of(
        "id", id,
        "backfillSupervisorId", backfillSupervisorId
    );
  }

  private void validateResetAndBackfill(
      String id,
      SeekableStreamSupervisor streamSupervisor,
      SeekableStreamSupervisorSpec streamSpec
  )
  {
    if (streamSupervisor.getIoConfig().isUseEarliestSequenceNumber()) {
      throw new IAE("Reset with skipped offsets is not supported when useEarliestOffset is true.");
    }

    if (!specHasConcurrentLocks(streamSpec)) {
      throw new IAE(
          "Backfill tasks require 'useConcurrentLocks' to be set to true in the supervisor context to allow concurrent writes with the main supervisor tasks"
      );
    }

    if (streamSupervisor.getState() != SupervisorStateManager.BasicState.RUNNING) {
      throw new IAE("Supervisor[%s] must be in a RUNNING state to perform a reset and backfill", id);
    }
  }

  public boolean checkPointDataSourceMetadata(
      String supervisorId,
      int taskGroupId,
      DataSourceMetadata previousDataSourceMetadata
  )

View on GitHub (pinned to 9b90983fd2)