transport.disconnect(true, true);
            } catch (Exception e) {
                log.debug("Error disconnecting unhealthy transport: {}", e.getMessage());
            }
        }

        if (!unhealthy.isEmpty()) {
            log.info("Removed {} unhealthy connections from pool", unhealthy.size());
        }

        // Perform maintenance tasks
        performMaintenanceTasks();
    }

    /**

		poolIDL, setIDL)
	erasureSetReadHealthMD = NewGaugeMD(erasureSetReadHealth,
		"Health of the erasure set in a pool for read operations (1=healthy, 0=unhealthy)",
		poolIDL, setIDL)
	erasureSetWriteHealthMD = NewGaugeMD(erasureSetWriteHealth,
		"Health of the erasure set in a pool for write operations (1=healthy, 0=unhealthy)",
		poolIDL, setIDL)
)

func b2f(v bool) float64 {
	if v {
		return 1
	}
	return 0
}

## Monitoring MinIO in Kubernetes

MinIO server exposes un-authenticated liveness endpoints so Kubernetes can natively identify unhealthy MinIO containers. MinIO also exposes Prometheus compatible data on a different endpoint to enable Prometheus users to natively monitor their MinIO deployments.

## Explore Further

              true
            }
          }
        }
      if (!acquired) continue

      // Confirm the connection is healthy and return it.
      if (connection.isHealthy(doExtensiveHealthChecks)) return connection

      // In the second synchronized block, release the unhealthy acquired connection. We're also on
      // the hook to close this connection if it's no longer in use.
      val noNewExchangesEvent: Boolean

| `minio_cluster_erasure_set_read_health`          | Health of the erasure set in a pool for read operations (1=healthy, 0=unhealthy). <br><br>Type: gauge   | `pool_id`, `set_id` |
| `minio_cluster_erasure_set_write_health`         | Health of the erasure set in a pool for write operations (1=healthy, 0=unhealthy). <br><br>Type: gauge  | `pool_id`, `set_id` |

#### `/cluster/health`

    // Make sure this connection is healthy & eligible for new exchanges. If it's no longer needed
    // then we're on the hook to close it.
    val healthy = candidate.isHealthy(connectionUser.doExtensiveHealthChecks())
    var noNewExchangesEvent = false
    val toClose: Socket? =
      candidate.withLock {
        when {
          !healthy -> {
            noNewExchangesEvent = !candidate.noNewExchanges

 * attempting to read with a short timeout. If the fails immediately we know the socket is
 * unhealthy.
 *
 * @param source the source used to read bytes from the socket.
 */
internal fun Socket.isHealthy(source: BufferedSource): Boolean =
  try {
    val readTimeout = soTimeout
    try {
      soTimeout = 1
      !source.exhausted()
    } finally {

        public void decrementConnections() {
            activeConnections.decrementAndGet();
        }

        public boolean isHealthy() {
            return isHealthy;
        }

        public void setHealthy(boolean healthy) {
            this.isHealthy = healthy;
            this.lastHealthCheck = System.currentTimeMillis();
        }

        public long getLastHealthCheck() {

    `okhttp-tls` module.

 *  Upgrade: [Kotlin 1.3.71][kotlin_1_3_71].


## Version 4.5.0

_2020-04-06_

**This release fixes a severe bug where OkHttp incorrectly detected and recovered from unhealthy
connections.** Stale or canceled connections were incorrectly attempted when they shouldn't have
been, leading to rare cases of infinite retries. Please upgrade to this release!

   * - The second request uses the coalesced connection from request1.
   * - The coalesced connection is violently closed after servicing the first request.
   * - The second request discovers the coalesced connection is unhealthy just after acquiring it.
   */
  @Test
  fun coalescedConnectionDestroyedAfterAcquire() {
    server.enqueue(MockResponse())
    server.enqueue(MockResponse())