```

##### Node (optional)

Optionally you can also collect per node metrics. This needs to be done on a per server instance.
The scrape configurations should use all the servers under `targets` so that graphing systems like
grafana can visualize them for all the nodes

```yaml
scrape_configs:
- job_name: minio-job
  metrics_path: /minio/v2/metrics/node
  scheme: http
  static_configs:

### Node Binaries

filename | sha256 hash
-------- | -----------
[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.5.8/kubernetes-node-linux-amd64.tar.gz) | `4d73ccd2ecac0f2e161f88e4d77004298d10a009f9b5fa0203fa7bff70a82e30`

```yaml
version: v2
address: ":9000"
rootUser: "minioadmin"
rootPassword: "minioadmin"
console-address: ":9001"
certs-dir: "/home/user/.minio/certs/"
pools: # Specify the nodes and drives with pools
  - args:
      - "https://server-example-pool1:9000/mnt/disk{1...4}/"
      - "https://server{1...2}-pool1:9000/mnt/disk{1...4}/"
      - "https://server3-pool1:9000/mnt/disk{1...4}/"

Kubernetes manages stateless Spark and Hive containers elastically on the compute nodes. Spark has native scheduler integration with Kubernetes. Hive, for legacy reasons, uses YARN scheduler on top of Kubernetes.

### Node Binaries

filename | sha512 hash
-------- | -----------
[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.26.15/kubernetes-node-linux-amd64.tar.gz) | a900a88f15e0a056c8673ae7e8931f5d91771dadc5841c3b04053f78efbb63c01d8589d0bfb70fc06fb32ac8816fccee9abe64f17af9f29498e1e5427f26656c

- This change adds a new alpha feature Node Declared Features, which includes:
  - A new `Node.Status.DeclaredFeatures` field for Kubelet to publish node-specific features.
  - A library in `component-helpers` for feature registration and inference.
  - A scheduler plugin (`NodeDeclaredFeatures`) scheduler plugin to match pods with nodes that provide their required features.

		prometheus.NewDesc(
			prometheus.BuildFQName(minioNamespace, "nodes", "online"),
			"Total number of MinIO nodes online",
			nil, nil),
		prometheus.GaugeValue,
		float64(nodesUp),
	)
	ch <- prometheus.MustNewConstMetric(
		prometheus.NewDesc(
			prometheus.BuildFQName(minioNamespace, "nodes", "offline"),
			"Total number of MinIO nodes offline",
			nil, nil),
		prometheus.GaugeValue,
		float64(nodesDown),

        assertThat(graph.degree(node)).isEqualTo(graph.adjacentNodes(node).size() + selfLoopCount);
        assertThat(graph.predecessors(node)).isEqualTo(graph.adjacentNodes(node));
        assertThat(graph.successors(node)).isEqualTo(graph.adjacentNodes(node));
        assertThat(graph.inDegree(node)).isEqualTo(graph.degree(node));
        assertThat(graph.outDegree(node)).isEqualTo(graph.degree(node));

      node = root.ceiling(comparator(), endpoint);
      if (node == null) {
        return null;
      }
      if (range.getLowerBoundType() == BoundType.OPEN
          && comparator().compare(endpoint, node.getElement()) == 0) {
        node = node.succ();
      }
    } else {
      node = header.succ();
    }
    return (node == header || !range.contains(node.getElement())) ? null : node;
  }

Example 1: Start distributed MinIO instance on n nodes with m drives each mounted at `/export1` to `/exportm` (pictured below), by running this command on all the n nodes:

![Distributed MinIO, n nodes with m drives each](https://github.com/minio/minio/blob/master/docs/screenshots/Architecture-diagram_distributed_nm.png?raw=true)

### GNU/Linux and macOS