apache/hadoop · error · java.io.IOException

It looks like the shared log is already being upgraded but t

Error message

It looks like the shared log is already being upgraded but this NN has not been upgraded yet. You should restart this NameNode with the '-bootstrapStandby' option to bring this NN in sync with the other.

What it means

Error "It looks like the shared log is already being upgraded but this NN has not been upgraded yet. You should restart this NameNode with the '-bootstrapStandby' option to bring this NN in sync with the other." thrown in apache/hadoop.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/FSImage.java:850

    Preconditions.checkState(getNamespaceID() != 0,
        "Must know namespace ID before initting edit log");
    String nameserviceId = DFSUtil.getNamenodeNameServiceId(conf);
    if (!HAUtil.isHAEnabled(conf, nameserviceId)) {
      // If this NN is not HA
      editLog.initJournalsForWrite();
      editLog.recoverUnclosedStreams();
    } else if (HAUtil.isHAEnabled(conf, nameserviceId)
        && (startOpt == StartupOption.UPGRADE
            || startOpt == StartupOption.UPGRADEONLY
            || RollingUpgradeStartupOption.ROLLBACK.matches(startOpt))) {
      // This NN is HA, but we're doing an upgrade or a rollback of rolling
      // upgrade so init the edit log for write.
      editLog.initJournalsForWrite();
      if (startOpt == StartupOption.UPGRADE
          || startOpt == StartupOption.UPGRADEONLY) {
        long sharedLogCTime = editLog.getSharedLogCTime();
        if (this.storage.getCTime() < sharedLogCTime) {
          throw new IOException("It looks like the shared log is already " +
              "being upgraded but this NN has not been upgraded yet. You " +
              "should restart this NameNode with the '" +
              StartupOption.BOOTSTRAPSTANDBY.getName() + "' option to bring " +
              "this NN in sync with the other.");
        }
      }
      editLog.recoverUnclosedStreams();
    } else {
      // This NN is HA and we're not doing an upgrade.
      editLog.initSharedJournalsForRead();
    }
  }

  /**
   * @param imageFile the image file that was loaded
   * @param numEditsLoaded the number of edits loaded from edits logs
   * @return true if the NameNode should automatically save the namespace
   * when it is started, due to the latest checkpoint being too old.

View on GitHub (pinned to 2add963021)

Solutions

1. Restart this NameNode with the '-bootstrapStandby' option to sync its storage with the upgraded active NameNode.

When it happens

Trigger: Starting a standby NameNode whose shared edits directory was already upgraded by the active NN while this NN still runs the older software.

Common situations: See trigger scenarios.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/5d3e18df572a5dcb. Report an issue: GitHub.