apache/hadoop · error · ConnectionNullException

Cannot get a connection to {}

Error message

Cannot get a connection to {}

What it means

Error "Cannot get a connection to {}" thrown in apache/hadoop.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RouterRpcClient.java:442

      // for each individual request.

      // TODO Add tokens from the federated UGI
      UserGroupInformation connUGI = ugi;
      if (UserGroupInformation.isSecurityEnabled() || this.enableProxyUser) {
        UserGroupInformation routerUser = UserGroupInformation.getLoginUser();
        connUGI = UserGroupInformation.createProxyUser(
            ugi.getUserName(), routerUser);
      }
      connection = this.connectionManager.getConnection(
          connUGI, rpcAddress, proto, nsId);
      LOG.debug("User {} NN {} is using connection {}",
          ugi.getUserName(), rpcAddress, connection);
    } catch (Exception ex) {
      LOG.error("Cannot open NN client to address: {}", rpcAddress, ex);
    }

    if (connection == null) {
      throw new ConnectionNullException("Cannot get a connection to "
          + rpcAddress);
    }
    return connection;
  }

  /**
   * Convert an exception to an IOException.
   *
   * For a non-IOException, wrap it with IOException. For a RemoteException,
   * unwrap it. For an IOException which is not a RemoteException, return it.
   *
   * @param e Exception to convert into an exception.
   * @return Created IO exception.
   */
  private static IOException toIOException(Exception e) {
    if (e instanceof RemoteException) {
      return ((RemoteException) e).unwrapRemoteException();
    }

View on GitHub (pinned to 2add963021)

Solutions

1. Verify the target namenode RPC address is correct and the namenode is running and reachable from the Router.
2. Check Kerberos credentials/token validity on the Router for the target nameservice.
3. Confirm network/firewall rules allow the Router to reach the namenode RPC port.

When it happens

Trigger: RouterRpcClient failed to create an RPC connection/proxy to the given namenode address.

Common situations: Namenode down, wrong RPC address in federation config, expired Kerberos ticket, or network partition between Router and a subcluster.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/e5350c77e7da20e2. Report an issue: GitHub.