})

			healthy := erasureSetUpCount[poolIdx][setIdx].online >= poolWriteQuorums[poolIdx]
			if !healthy {
				storageLogIf(logger.SetReqInfo(ctx, reqInfo),
					fmt.Errorf("Write quorum may be lost on pool: %d, set: %d, expected write quorum: %d",
						poolIdx, setIdx, poolWriteQuorums[poolIdx]), logger.FatalKind)
			}
			result.Healthy = result.Healthy && healthy

		"Overall write quorum across pools and sets")
	erasureSetOverallHealthMD = NewGaugeMD(erasureSetOverallHealth,
		"Overall health across pools and sets (1=healthy, 0=unhealthy)")
	erasureSetReadQuorumMD = NewGaugeMD(erasureSetReadQuorum,
		"Read quorum for the erasure set in a pool", poolIDL, setIDL)
	erasureSetWriteQuorumMD = NewGaugeMD(erasureSetWriteQuorum,

| `minio_cluster_erasure_set_overall_write_quorum` | `gauge` | Overall write quorum across pools and sets                    |                  |
| `minio_cluster_erasure_set_overall_health`       | `gauge` | Overall health across pools and sets (1=healthy, 0=unhealthy) |                  |
| `minio_cluster_erasure_set_read_quorum`          | `gauge` | Read quorum for the erasure set in a pool                     | `pool_id,set_id` |

		opts := HealthOptions{}
		result := objLayer.Health(ctx, opts)

		metrics = make([]MetricV2, 0, 2+4*len(result.ESHealth))

		metrics = append(metrics, MetricV2{
			Description: getClusterWriteQuorumMD(),
			Value:       float64(result.WriteQuorum),
		})

		health := 1
		if !result.Healthy {
			health = 0
		}

		metrics = append(metrics, MetricV2{

		return ctx, done, ctx.Err()
	}

	atomic.StoreInt64(&p.health.lastStarted, time.Now().UnixNano())
	p.health.waiting.Add(1)

	ctx = context.WithValue(ctx, healthDiskCtxKey{}, &healthDiskCtxValue{lastSuccess: &p.health.lastSuccess})
	si := p.updateStorageMetrics(s, paths...)
	var once sync.Once
	return ctx, func(errp *error) {
		p.health.waiting.Add(-1)
		once.Do(func() {
			if errp != nil {
				err := *errp

		Online:   true,
	}
	sys.hMutex.Unlock()
}

// newHCClient initializes an anonymous client for performing health check on the remote endpoints
func newHCClient() *madmin.AnonymousClient {
	clnt, e := madmin.NewAnonymousClientNoEndpoint()
	if e != nil {
		bugLogIf(GlobalContext, errors.New("Unable to initialize health check client"))
		return nil
	}
	clnt.SetCustomTransport(globalRemoteTargetTransport)
	return clnt
}

          "expr": "minio_cluster_health_status{job=~\"$scrape_jobs\"}",
          "fullMetaSearch": false,
          "includeNullMetadata": true,
          "interval": "",
          "legendFormat": "Pool: {{pool}} Set: {{set}}",
          "range": true,
          "refId": "A",
          "useBackend": false
        }
      ],
      "title": "Cluster Health Status",
      "type": "stat"
    },
    {

	}
	result := objLayer.Health(ctx, opts)
	w.Header().Set(xhttp.MinIOWriteQuorum, strconv.Itoa(result.WriteQuorum))
	w.Header().Set(xhttp.MinIOStorageClassDefaults, strconv.FormatBool(result.UsingDefaults))
	// return how many drives are being healed if any
	if result.HealingDrives > 0 {
		w.Header().Set(xhttp.MinIOHealingDrives, strconv.Itoa(result.HealingDrives))
	}
	if !result.Healthy {

		"Count of offline nodes in the cluster")
	healthNodesOnlineCountMD = NewGaugeMD(healthNodesOnlineCount,
		"Count of online nodes in the cluster")
)

// loadClusterHealthNodeMetrics - `MetricsLoaderFn` for cluster health node
// metrics.
func loadClusterHealthNodeMetrics(ctx context.Context, m MetricValues,
	c *metricsCache,
) error {
	nodesUpDown, _ := c.nodesUpDown.Get()

	m.Set(healthNodesOfflineCount, float64(nodesUpDown.Offline))

## Liveness probe

This probe always responds with '200 OK'. Only fails if 'etcd' is configured and unreachable. When liveness probe fails, Kubernetes like platforms restart the container.

```
livenessProbe:
  httpGet:
    path: /minio/health/live
    port: 9000
    scheme: HTTP
  initialDelaySeconds: 120