var (
		// We check every 15 seconds if the disk is writable and we can read back.
		checkEvery = 15 * time.Second

		// If the disk has completed an operation successfully within last 5 seconds, don't check it.
		skipIfSuccessBefore = 5 * time.Second
	)

	// if disk max timeout is smaller than checkEvery window
	// reduce checks by a second.
	if globalDriveConfig.GetMaxTimeout() <= checkEvery {

C'est assez simple (pas très utile), mais cela nous aidera à nous concentrer sur le fonctionnement des sous-dépendances.

## Créer une seconde dépendance, « dependable » et « dependant » { #second-dependency-dependable-and-dependant }

Vous pouvez ensuite créer une autre fonction de dépendance (un « dependable ») qui, en même temps, déclare sa propre dépendance (elle est donc aussi un « dependant ») :

        // Need successThreshold (2) consecutive successes to close the circuit
        // Second success should close the circuit (threshold is 2)
        result = circuitBreaker.executeWithCircuitBreaker(() -> "success 1");
        assertEquals("success 1", result, "Should execute second success");
        assertEquals(State.CLOSED, circuitBreaker.getState(), "Circuit should be CLOSED after success threshold");

    @Keep public static final WithPublicConstants SECOND = new WithPublicConstants();
  }

  private static class FirstConstantIsNull {
    // To test that null constant is ignored
    @Keep public static final @Nullable FirstConstantIsNull FIRST = null;

    @Keep public static final FirstConstantIsNull SECOND = new FirstConstantIsNull();
  }

  public static class NonFinalFieldIgnored {

     */
    private long determinePollInterval(ChangeNotificationHandle handle) {
        int failures = getFailureCount(handle);

        // Base interval starts at 1 second, increases with failures
        // Max interval is 30 seconds for inactive directories
        return Math.min(MAX_POLL_INTERVAL, BASE_POLL_INTERVAL + (failures * 2000));
    }

    /**
     * Get failure count for a handle
     *

        byte[] nonce2 = context.generateSecureNonce();

        // Then
        assertNotNull(nonce1, "First secure nonce should not be null");
        assertNotNull(nonce2, "Second secure nonce should not be null");
        assertFalse(Arrays.equals(nonce1, nonce2), "Secure nonces should be different");
        assertTrue(nonce1.length > 0, "Secure nonce should have proper length");
    }

  override fun send(text: String): Boolean = send(text.encodeUtf8(), OPCODE_TEXT)

  override fun send(bytes: ByteString): Boolean = send(bytes, OPCODE_BINARY)

  @Synchronized private fun send(
    data: ByteString,
    formatOpcode: Int,
  ): Boolean {
    // Don't send new frames after we've failed or enqueued a close frame.

    }

    /**
     * Returns the second value.
     *
     * @return The second value
     */
    public T2 getSecond() {
        return second;
    }

    /**
     * Sets the second value.
     *
     * @param second
     *            The second value
     */
    public void setSecond(final T2 second) {
        this.second = second;
    }

    @Override

> Notice the requirement of common SLA here original cluster had 1024 drives with 16 drives per erasure set with default parity of '4', second pool is expected to have a minimum of 8 drives per erasure set to match the original cluster SLA (parity count) of '4'. '12' drives stripe per erasure set in the second pool satisfies the original pool's parity count.

     *
     * Operations larger than this size will use RDMA read/write
     * instead of send/receive.
     *
     * @return threshold in bytes
     */
    int getRdmaReadWriteThreshold();

    /**
     * Get maximum RDMA send size
     *
     * @return max send size in bytes
     */
    int getRdmaMaxSendSize();

    /**
     * Get maximum RDMA receive size
     *