What causes "failed to close one or more emitters" in apache/druid?

Check the accompanying 'Failed to close emitter [ ]' log lines to identify which child emitter(s) failed and fix the root cause (network, endpoint availability).

apache/druid · error · IOException

failed to close one or more emitters

Error message

failed to close one or more emitters

What it means

SwitchingEmitter routes events to one of several underlying emitters (alert, feed, etc.). During close(), it closes each child emitter; if any child throws an IOException, it remembers the failure and throws this aggregate IOException once all children have been attempted. It signals that at least one child emitter could not shut down cleanly, so buffered events may not have been flushed.

Source

Thrown at processing/src/main/java/org/apache/druid/java/util/emitter/core/SwitchingEmitter.java:164

  @LifecycleStop
  public void close() throws IOException
  {
    boolean fail = false;
    log.info("Closing Switching Emitter.");

    for (Emitter e : knownEmitters) {
      try {
        log.info("Closing emitter %s.", e.getClass().getName());
        e.close();
      }
      catch (IOException ex) {
        log.error(ex, "Failed to close emitter [%s]", e.getClass().getName());
        fail = true;
      }
    }

    if (fail) {
      throw new IOException("failed to close one or more emitters");
    }
  }
}

View on GitHub (pinned to 9b90983fd2)

Solutions

Check the accompanying 'Failed to close emitter [<class>]' log lines to identify which child emitter(s) failed and fix the root cause (network, endpoint availability).
Ensure the child emitters' target endpoints are reachable and draining correctly before/while shutting down the service.
Verify close() is only called once per emitter lifecycle; avoid closing an already-closed emitter.
If the failure is during process shutdown and data loss is acceptable for those metrics, this can often be logged and ignored.
Flush or wait for the emitter's internal queue to drain (e.g. call flush() and wait) before close() to reduce failure risk.

Example fix

// before
switchingEmitter.close(); // may throw IOException, swallowing other pending shutdown work
// after
try {
  emitter.flush();
  switchingEmitter.close();
} catch (IOException e) {
  log.warn(e, "Some emitters failed to close during shutdown; continuing");
}

Defensive patterns

Strategy: try-catch

Validate before calling

// before close, ensure children can flush
switchingEmitter.flush();

Try / catch

try {
  switchingEmitter.close();
} catch (IOException e) {
  log.warn(e, "Emitter close failed on shutdown; some events may not have been flushed");
}

Prevention

Register emitters in the Druid lifecycle so close happens once, in order, at shutdown
Keep emitter endpoints healthy; monitor the target service availability
Call flush() and let queues drain before close()
Never close the same emitter twice

When it happens

Trigger: Calling SwitchingEmitter.close() when any wrapped emitter's close() throws IOException — typically a UriEmitter/HttpEmitter whose flush to its HTTP endpoint fails, or an emitter whose writer is already closed or errored.

Common situations: Druid service shutdown while the HTTP emitter's target (e.g. a metrics/alerting endpoint) is down or slow; double-close of emitters; disk or network issues preventing the child emitter from flushing its queue during close.

Understand the failure class

Background: "failed to write file", "Could not save figure", "Error saving remote file" — file write failed: causes and fixes across languages and libraries — this error's family across 38 libraries.

Related errors

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/5d2b4d9048435924. Report an issue: GitHub.