apache/hadoop · error · IOException

Invalid reservationId: {} specified for the app: {}

Error message

Invalid reservationId: {} specified for the app: {}

What it means

YARNRunner.submitJob reads mapreduce.job.reservation.id and passes it to ReservationId.parseReservationId. That method accepts 'reservation_<clusterTimestamp>_<seq>' and, when the prefix and field count pass but the timestamp or sequence part is not numeric, throws NumberFormatException. YARNRunner converts it into an IOException that names the bad value and the application ID, aborting job submission before the ApplicationSubmissionContext reaches the ResourceManager.

Source

Thrown at hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/main/java/org/apache/hadoop/mapred/YARNRunner.java:622

    ApplicationSubmissionContext appContext =
        recordFactory.newRecordInstance(ApplicationSubmissionContext.class);
    appContext.setApplicationId(applicationId);                // ApplicationId
    appContext.setQueue(                                       // Queue name
        jobConf.get(JobContext.QUEUE_NAME,
        YarnConfiguration.DEFAULT_QUEUE_NAME));
    // add reservationID if present
    ReservationId reservationID = null;
    try {
      reservationID =
          ReservationId.parseReservationId(jobConf
              .get(JobContext.RESERVATION_ID));
    } catch (NumberFormatException e) {
      // throw exception as reservationid as is invalid
      String errMsg =
          "Invalid reservationId: " + jobConf.get(JobContext.RESERVATION_ID)
              + " specified for the app: " + applicationId;
      LOG.warn(errMsg);
      throw new IOException(errMsg);
    }
    if (reservationID != null) {
      appContext.setReservationID(reservationID);
      LOG.info("SUBMITTING ApplicationSubmissionContext app:" + applicationId
          + " to queue:" + appContext.getQueue() + " with reservationId:"
          + appContext.getReservationID());
    }
    appContext.setApplicationName(                             // Job name
        jobConf.get(JobContext.JOB_NAME,
        YarnConfiguration.DEFAULT_APPLICATION_NAME));
    appContext.setCancelTokensWhenComplete(
        conf.getBoolean(MRJobConfig.JOB_CANCEL_DELEGATION_TOKEN, true));
    appContext.setAMContainerSpec(amContainer);         // AM Container
    appContext.setMaxAppAttempts(
        conf.getInt(MRJobConfig.MR_AM_MAX_ATTEMPTS,
            MRJobConfig.DEFAULT_MR_AM_MAX_ATTEMPTS));

    // Setup the AM ResourceRequests

View on GitHub (pinned to 2add963021)

Solutions

1. Use the exact string returned by ReservationId.toString() from the reservation system that created it (format: reservation_<clusterTimestamp>_<sequence>, both numeric)
2. If no reservation is intended, unset mapreduce.job.reservation.id instead of leaving a placeholder value
3. Validate the value with ReservationId.parseReservationId before job submission so the failure surfaces at the configuration boundary, not inside submitJob
4. Check the ResourceManager/capacity scheduler reservationACLs and reservation system logs to re-obtain the correct ID if it came from an automation pipeline

Example fix

// before
conf.set("mapreduce.job.reservation.id", "res-2026-07-01");

// after
ReservationId rid = reservationClient.getReservationReservationId(...); // from ReservationSystem
conf.set("mapreduce.job.reservation.id", rid.toString()); // e.g. "reservation_1689372000000_0001"

String rid = conf.get(MRJobConfig.RESERVATION_ID); // mapreduce.job.reservation.id
if (rid != null) {
  try {
    ReservationId.parseReservationId(rid); // expects reservation_<ts>_<seq>, both numeric
  } catch (IOException | NumberFormatException e) {
    throw new IllegalArgumentException("Refusing to submit: bad reservation id: " + rid, e);
  }
}

try {
  RunningJob job = jobClient.submitJob(jobConf);
} catch (IOException e) {
  if (e.getMessage() != null && e.getMessage().startsWith("Invalid reservationId:")) {
    // re-read the reservation from the ReservationSystem and resubmit
  }
  throw e;
}

When it happens

Trigger: Setting mapreduce.job.reservation.id (JobContext.RESERVATION_ID) to a string like 'reservation_abc_1' or 'reservation_1234_xyz' (non-numeric timestamp/sequence fields), then calling JobClient.submitJob / YARNRunner.submitJob. Note: values with a wrong prefix or wrong underscore-field count fail inside parseReservationId with IOException before the NumberFormatException path, so this specific message implies the shape was right but a field was non-numeric.

Common situations: Hand-typed reservation IDs in scripts or CLI flags; reservation IDs copied with extra characters or split across shell variables; passing IDs generated by a different cluster or tool that formats timestamps with separators; forgetting that only IDs returned by the YARN ReservationSystem are valid.

Related errors

• "Not creating intermediate history logDir: [" + doneDirPath
• ShuffleProvider Service: {} was NOT found in the list of aux
• Invalid specification for distributed-cache artifacts of typ
• {} is incompatible with {} mode.
• Not submitting job. Job directory {} already exists!! This i

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/df7411e8b859a868. Report an issue: GitHub.