apache/hadoop · error · IOException

Invalid timeout [timeout = {connectionTimeout} ms]

Error message

Invalid timeout [timeout = {connectionTimeout} ms]

What it means

Fetcher.connect() slices the configured connect timeout into unit attempts (each capped at UNIT_CONNECT_TIMEOUT) and retries until the total budget is spent. A negative mapreduce.reduce.shuffle.connect.timeout cannot be sliced, so every connect attempt of the task fails immediately with this IOException.

Source

Thrown at hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/task/reduce/Fetcher.java:706

      url.append(mapId);
      first = false;
    }

    if (LOG.isDebugEnabled()) {
      LOG.debug("MapOutput URL for " + host + " -> " + url.toString());
    }
    return new URL(url.toString());
  }
  
  /** 
   * The connection establishment is attempted multiple times and is given up 
   * only on the last failure. Instead of connecting with a timeout of 
   * X, we try connecting with a timeout of x < X but multiple times. 
   */
  private void connect(URLConnection connection, int connectionTimeout) throws IOException {
    int unit = 0;
    if (connectionTimeout < 0) {
      throw new IOException("Invalid timeout "
                            + "[timeout = " + connectionTimeout + " ms]");
    } else if (connectionTimeout > 0) {
      unit = Math.min(UNIT_CONNECT_TIMEOUT, connectionTimeout);
    }
    long startTime = Time.monotonicNow();
    long lastTime = startTime;
    int attempts = 0;
    // set the connect timeout to the unit-connect-timeout
    connection.setConnectTimeout(unit);
    while (true) {
      try {
        attempts++;
        connection.connect();
        break;
      } catch (IOException ioe) {
        long currentTime = Time.monotonicNow();
        long retryTime = currentTime - startTime;
        long leftTime = connectionTimeout - retryTime;

View on GitHub (pinned to 2add963021)

Solutions

1. Set mapreduce.reduce.shuffle.connect.timeout to a positive millisecond value (default 130000).
2. Remove the override entirely to fall back to the default.
3. If the intent was 'no timeout', use a large positive value such as 600000 instead of -1.

Example fix

// before
conf.setInt("mapreduce.reduce.shuffle.connect.timeout", -1);
// after
conf.setInt("mapreduce.reduce.shuffle.connect.timeout", 130000);

// Validate shuffle timeouts before job submission
int connectTimeout = conf.getInt("mapreduce.reduce.shuffle.connect.timeout", 130000);
int readTimeout = conf.getInt("mapreduce.reduce.shuffle.read.timeout", 80000);
if (connectTimeout <= 0 || readTimeout <= 0) { throw new IllegalArgumentException("Shuffle timeouts must be positive milliseconds: connect=" + connectTimeout + ", read=" + readTimeout); }

When it happens

Trigger: Job configuration sets mapreduce.reduce.shuffle.connect.timeout to a negative value, typically -1 by convention from other systems where it means 'infinite'.

Common situations: Copying timeout settings from other frameworks (curl, netty, postgres) where -1 disables the timeout; templates or variable substitution producing -1; misparsed integer in the job xml.

Understand the failure class

• Timeouts: ETIMEDOUT, deadlines, and hung requests — what actually expires when a request times out.

Related errors

• Invalid value for mapreduce.reduce.shuffle.input.buffer.perc
• Invalid value for mapreduce.reduce.shuffle.memory.limit.perc
• Invalid configuration: maxSingleShuffleLimit should be less
• {maxRedPer}: mapreduce.reduce.input.buffer.percent must be a
• Malformed xml formation queue tag and acls tags or state ta

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/bd5c62af5bd52e0d. Report an issue: GitHub.