# Compression Guide [![Slack](https://slack.min.io/slack?type=svg)](https://slack.min.io)

MinIO server allows streaming compression to ensure efficient disk space usage.
Compression happens inflight, i.e objects are compressed before being written to disk(s).
MinIO uses [`klauspost/compress/s2`](https://github.com/klauspost/compress/tree/master/s2)
streaming compression due to its stability and performance.

// in_flight_nodes_limit: when is_async is true, this value controls the
// maximum number of in flight async nodes. Enqueuing of additional async ops
// after the limit is reached blocks until some inflight nodes finishes.
// The effect is bounding the memory held by inflight TensorHandles that are
// referenced by the inflight nodes.
// A recommended value has not been established.

          "interval": "1m",
          "intervalFactor": 2,
          "legendFormat": "{{bucket,api}}",
          "refId": "A"
        }
      ],
      "title": "Inflight Requests",
      "type": "timeseries"
    },
    {
      "datasource": {
        "type": "prometheus",
        "uid": "${DS_PROMETHEUS}"
      },
      "fieldConfig": {
        "defaults": {

* Add apiserver metric for number of requests dropped because of inflight limit. ([#58340](https://github.com/kubernetes/kubernetes/pull/58340), [@gmarek](https://github.com/gmarek))
* Add apiserver metric for current inflight-request usage. ([#58342](https://github.com/kubernetes/kubernetes/pull/58342), [@gmarek](https://github.com/gmarek))

   * The maximum number of requests to execute concurrently. Above this requests queue in memory,
   * waiting for the running calls to complete.
   *
   * If more than [maxRequests] requests are in flight when this is invoked, those requests will
   * remain in flight.
   */
  var maxRequests = 64
    get() = this.withLock { field }
    set(maxRequests) {
      require(maxRequests >= 1) { "max < 1: $maxRequests" }
      this.withLock {

	apiRequestsIncomingTotalMD = NewGaugeMD(apiRequestsIncomingTotal,
		"Total number of incoming requests", "type")

	apiRequestsInFlightTotalMD = NewGaugeMD(apiRequestsInFlightTotal,
		"Total number of requests currently in flight", "name", "type")
	apiRequestsTotalMD = NewCounterMD(apiRequestsTotal,
		"Total number of requests", "name", "type")
	apiRequestsErrorsTotalMD = NewCounterMD(apiRequestsErrorsTotal,

 *     lookups) and attempt a new connection them. When failures occur, retries iterate the list of
 *     available routes.
 *
 * If the pool gains an eligible connection while DNS, TCP, or TLS work is in flight, this finder
 * will prefer pooled connections. Only pooled HTTP/2 connections are used for such de-duplication.
 *
 * It is possible to cancel the finding process by canceling its call.
 *

	mlock sync.RWMutex // mutex for bucket measurement

	bucketsThrottle    map[BucketOptions]*bucketThrottle
	bucketsMeasurement map[BucketOptions]*bucketMeasurement // Buckets with objects in flight

	bucketMovingAvgTicker *time.Ticker    // Ticker for calculating moving averages
	ctx                   context.Context // Context for generate
	NodeCount             uint64
}

          connectResult = launchTcpConnect()
          nextTcpConnectAtNanos = now + connectDelayNanos
          awaitTimeoutNanos = connectDelayNanos
        }

        // Wait for an in-flight connect to complete or fail.
        if (connectResult == null) {
          connectResult = awaitTcpConnect(awaitTimeoutNanos, TimeUnit.NANOSECONDS) ?: continue
        }

        if (connectResult.isSuccess) {

    )

    taskFaker.assertNoMoreTasks()
  }

  /**
   * This test puts several connections in flight that all fail at approximately the same time. It
   * confirms the fast fallback implements these invariants:
   *
   *  * if there's no TCP connect in flight, start one.
   *  * don't start a new TCP connect within 250 ms of the previous TCP connect.
   */
  @Test