> For the complete documentation index, see [llms.txt](https://documentation.alluxio.io/ee-ai-en/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://documentation.alluxio.io/ee-ai-en/administration.md).

# Administration

This section provides a high-level overview of administering an Alluxio cluster, covering key areas from day-to-day management and monitoring to security and troubleshooting.

## 1. Managing the Cluster

Cluster administration is split across several focused pages. Use the table below to find the right page for the task at hand.

| Question / Task                                                                       | Page                                                                                                                                                                                        |
| ------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Harden a basic install for production (node pinning, HA, resource tuning)             | [Cluster Management — Production Setup](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#production-setup)                                                      |
| Scale the cluster up or down                                                          | [Cluster Management — Scaling](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#scaling-the-cluster)                                                            |
| Upgrade Alluxio to a new version                                                      | [Cluster Management — Upgrading](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#upgrading-alluxio)                                                            |
| Change a property on a running cluster without a restart                              | [Dynamic Configuration](https://documentation.alluxio.io/ee-ai-en/administration/dynamic-configuration.md)                                                                                  |
| Change any other property (restart needed)                                            | [Cluster Management — Updating Configuration](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#updating-configuration)                                          |
| Tune the consistent hash ring (mode, virtual nodes, capacity)                         | [Cluster and Membership Management — Hash Ring Configuration](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#hash-ring-configuration)                         |
| Diagnose stale `OFFLINE` workers / hash ring bloat                                    | [Cluster and Membership Management — Hash Ring Bloat](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#diagnosing-hash-ring-bloat-from-offline-entries)         |
| Add, remove, restart, or persist identity for a worker                                | [Cluster and Membership Management — Worker Membership and Lifecycle](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#worker-membership-and-lifecycle)         |
| Configure worker page store (hostPath / PVC, sizing, multi-disk)                      | [Worker Configuration — Worker Storage](https://documentation.alluxio.io/pages/MF8PfNU0P7q8PD9eh6xW#id-1.-worker-storage)                                                                   |
| Set up heterogeneous workers                                                          | [Worker Configuration — Heterogeneous Workers](https://documentation.alluxio.io/ee-ai-en/administration/managing-worker.md#heterogeneous-workers)                                           |
| A page store or metastore disk failed, or a worker reports a directory out of service | [Disk Health — Recover from a Disk Failure](https://documentation.alluxio.io/ee-ai-en/administration/disk-health.md#recover-from-a-disk-failure)                                            |
| Alert on failing worker disks                                                         | [Disk Health — Apply the Alert Rules](https://documentation.alluxio.io/ee-ai-en/administration/disk-health.md#apply-the-alert-rules)                                                        |
| Tune worker resources / JVM, diagnose OOM                                             | [Worker Configuration — Resource and JVM Tuning](https://documentation.alluxio.io/pages/MF8PfNU0P7q8PD9eh6xW#id-2.-resource-and-jvm-tuning)                                                 |
| Bind a worker to a specific NIC                                                       | [Worker Configuration — Network](https://documentation.alluxio.io/pages/MF8PfNU0P7q8PD9eh6xW#id-3.-worker-network-configuration)                                                            |
| Tune how long clients wait on a slow or unreachable worker                            | [Worker Configuration — Client–Worker Timeouts](https://documentation.alluxio.io/pages/MF8PfNU0P7q8PD9eh6xW#id-4.-clientworker-timeouts)                                                    |
| Recover cache coverage after a worker crash or restart                                | [Cluster and Membership Management — Cache Recovery After Worker Restart](https://documentation.alluxio.io/ee-ai-en/administration/managing-alluxio.md#cache-recovery-after-worker-restart) |
| Rebalance cached data after adding workers                                            | [Cache Rebalancing](https://documentation.alluxio.io/ee-ai-en/cache/cache-rebalancing.md)                                                                                                   |
| Scale in or out without cache misses (rebalance cached data to the new topology)      | [Cache Rebalancing](https://documentation.alluxio.io/ee-ai-en/cache/cache-rebalancing.md)                                                                                                   |
| Job submission, scheduling, lifecycle, HA, recovery                                   | [Job Service](https://documentation.alluxio.io/ee-ai-en/administration/managing-job-service.md)                                                                                             |
| Point Alluxio at an etcd you manage, or tune the bundled one                          | [etcd Management](https://documentation.alluxio.io/ee-ai-en/administration/managing-etcd.md)                                                                                                |
| Move to a different etcd without losing the mount table or the cache                  | [etcd Management — Migrating](https://documentation.alluxio.io/ee-ai-en/administration/managing-etcd.md#migrating-to-a-different-etcd)                                                      |
| Tune load throughput for small files or large directories                             | [Job Service — Small File Optimization](https://documentation.alluxio.io/ee-ai-en/administration/managing-job-service.md#small-file-optimization)                                           |
| Multi-tenancy isolation and cluster federation                                        | [Cluster Management — Multi-Tenancy](https://documentation.alluxio.io/pages/Zi86aQPzWMrc2orY9tBY#id-3.-multi-tenancy-and-federation)                                                        |

## 2. Monitoring and Observability

Alluxio exposes extensive metrics in the Prometheus format, enabling deep visibility into the cluster's health and performance.

* **Default Monitoring Stack**: The Alluxio Operator can automatically deploy a complete monitoring stack, including Prometheus for metrics collection and Grafana for visualization with pre-configured dashboards.
* **Integration with Existing Systems**: You can easily integrate Alluxio with your existing monitoring infrastructure, whether it's a central Prometheus, Grafana, or a third-party service like Datadog.

Learn more about [Monitoring Alluxio](https://documentation.alluxio.io/ee-ai-en/administration/monitoring-alluxio.md)...

For interactive investigation, the coordinator can also serve a browser-based **Operations Console**: worker health, cache contents and coverage, read behavior including slow and failed reads, and per-worker JVM and host diagnostics.

Learn more about the [Operations Console](https://documentation.alluxio.io/ee-ai-en/administration/operations-console.md)...

## 3. Security

Alluxio provides a multi-layered security model to protect your data and infrastructure.

* **Authentication**: Secure your cluster by integrating with an OIDC-compliant Identity Provider (like Okta) to authenticate users and services using JSON Web Tokens (JWTs).
* **Authorization**: Enforce fine-grained access control. Use **Apache Ranger** for data access policies (S3, HDFS) and **Open Policy Agent (OPA)** for management API policies (Gateway).
* **Encryption**: Protect data in transit by enabling TLS to encrypt communication between Alluxio components and between clients and the cluster.
* **Audit Log**: Record management operations (Gateway) and data access (S3, HDFS, FUSE, Python SDK) as structured JSON for compliance and forensics.

Learn more about [Security](https://documentation.alluxio.io/ee-ai-en/administration/security.md)...

## 4. Troubleshooting

When issues arise, Alluxio provides tools and procedures to help you diagnose and resolve them quickly.

* **Health Checks**: Start by checking the status of Alluxio components (Coordinators, Workers, FUSE) and verifying connectivity to the UFS.
* **Diagnostics**: Inspect logs from Alluxio processes and Kubernetes CSI drivers. For complex issues, collect a [diagnostic bundle](https://documentation.alluxio.io/ee-ai-en/administration/diagnostic-bundle.md) with `alluxioctl doctor bundle`, which packs logs, configuration, state, and metrics into one file for offline analysis.
* **Recovery**: Follow guided procedures to recover from common failures, such as a failed coordinator, worker, or a corrupted etcd cluster.

Learn more about [Troubleshooting Alluxio](https://documentation.alluxio.io/ee-ai-en/administration/troubleshooting-alluxio.md)...
