- Failed pods associated with a job with `parallelism = 1` are recreated by the job controller honoring exponential backoff delay again. However, for jobs with `parallelism > 1`, pods might be created without exponential backoff delay. ([#115020](https://github.com/kubernetes/kubernetes/pull/115020), [@nikhita](https://github.com/nikhita)) [SIG Apps]

    sink.writeUtf8(data)
    sink.close()
    return result
  }

  private fun cancelLater(
    call: Call,
    delay: Long,
  ): Thread {
    val thread =
      object : Thread("canceler") {
        override fun run() {
          try {
            sleep(delay)
          } catch (e: InterruptedException) {
            throw AssertionError()
          }
          call.cancel()

    }
  }

  static final Ticker NULL_TICKER =
      new Ticker() {
        @Override
        public long read() {
          return 0;
        }
      };

  // We use a holder class to delay initialization: https://github.com/google/guava/issues/6566
  private static final class LoggerHolder {
    static final Logger logger = Logger.getLogger(CacheBuilder.class.getName());
  }

    intend to ship a public API for these features, but we don't want to hold OkHttp 5.0.0 until
    those APIs are stable.

 *  Fix: Change `MockWebServer.close()` to cancel ongoing calls that are blocked on a delay.

 *  Upgrade: [Okio 3.13.0][okio_3_13_0].

This release also stabilizes many APIs in the `mockwebserver3` artifact that's new in 5.0.

    public synchronized void close() {
        final SmbTreeHandleImpl th = this.treeHandle;
        if (th != null) {
            this.treeHandle = null;

            // Optimization: For non-strict lifecycle, delay tree handle closure
            // to allow connection reuse for subsequent operations
            if (this.transportContext.getConfig().isStrictResourceLifecycle()) {
                th.close();
            } else {

- Fixed computing backoff delay when using Job pod failure policy, by including in the backoff delay calculation pod failures ignored from the backoffLimit counter. ([#119434](https://github.com/kubernetes/kubernetes/pull/119434), [@mimowo](https://github.com/mimowo))

        // Node is back online - could rebalance connections
    }

    /**
     * Schedule a reconnection attempt
     *
     * @param delayMs delay in milliseconds
     */
    private void scheduleReconnection(long delayMs) {
        CompletableFuture.delayedExecutor(delayMs, TimeUnit.MILLISECONDS).execute(() -> {
            try {

Entonces podrías querer tener **un solo contenedor** con un **gestor de procesos** iniciando **varios worker processes** dentro.

---

El punto principal es que, **ninguna** de estas son **reglas escritas en piedra** que debas seguir a ciegas. Puedes usar estas ideas para **evaluar tu propio caso de uso** y decidir cuál es el mejor enfoque para tu sistema, verificando cómo gestionar los conceptos de:

* Seguridad - HTTPS
* Ejecutar en el inicio

- Fix computing backoff delay when using Job pod failure policy, by including in the backoff delay calculation pod failures ignored from the backoffLimit counter.
  
  Also, compute the backoff delay more accurately for deleted pods. ([#119466](https://github.com/kubernetes/kubernetes/pull/119466), [@mimowo](https://github.com/mimowo)) [SIG Apps]

- Fix: don't use docker config cache if it's empty ([#92330](https://github.com/kubernetes/kubernetes/pull/92330), [@andyzhangx](https://github.com/andyzhangx)) [SIG Cloud Provider]
- Fix: initial delay in mounting azure disk & file ([#93052](https://github.com/kubernetes/kubernetes/pull/93052), [@andyzhangx](https://github.com/andyzhangx)) [SIG Cloud Provider and Storage]