This release enhances cluster management and operational performance. Administrators gain finer control with customizable node health check intervals and new replication metrics, while performance is boosted for multipart uploads and high-load IAM operations. Critical fixes improve data management reliability in distributed environments and restore key administrative API endpoints.

### New Features

*   **Customizable Node Health Checks**: Administrators can now configure the health check interval in distributed deployments, enabling faster detection of unresponsive nodes to reduce failover time. (#851)
*   **Enhanced Replication Metrics**: Introduced new metrics for monitoring in-progress data replication tasks, providing a clearer view of replication health and status across all nodes. (#845)

### Improvements

*   **Expanded Elasticsearch Compatibility**: Upgraded event notification support to include Elasticsearch server versions 7, 8, and 9, ensuring seamless integration with modern logging and analytics deployments. (#820)
*   **Optimized IAM Performance**: Reduced processing delays for identity and access management (IAM) key checks under high load, improving server responsiveness in large-scale environments. (#863)
*   **Improved Multipart Upload Performance**: Enabled multipart upload caching (`mpCache`) by default to enhance performance for most standard deployments. (#866)
*   **Accurate Drive Health Monitoring**: Improved drive health check accuracy by independently monitoring read and write operations, ensuring drives with partial failures are correctly identified. (#848)
*   **Resilient Lifecycle Management**: The object lifecycle management (ILM) system now ignores "object not found" errors during expiration, preventing unnecessary log noise and ensuring smoother operation. (#855)
*   **Robust LDAP Configuration**: Implemented stricter validation for LDAP settings upon application, preventing misconfigurations and potential authentication issues. (#776)

### Bug Fixes

*   **Reliable Large File Uploads**: Resolved an issue causing '503 Service Unavailable' errors during large multipart uploads in high-latency, stretched cluster environments. (#861)
*   **Consistent Non-Versioned Deletes**: Ensured successful deletion of non-versioned objects even when a storage node is offline and metadata lacks quorum, preventing 503 errors. (#854)
*   **Bucket Quota Enforcement**: Fixed an issue to ensure storage quotas on buckets created with older software versions are correctly recognized and enforced after an upgrade. (#870)
*   **Restored Administrative API Endpoints**:
    *   Reinstated the `/minio/admin/v4/object-summary` endpoint to restore object summary retrieval for administrative tools. (#859)
    *   Reinstated the `/idp/openid/list-access-keys-bulk` and `/info-access-key` endpoints to restore OIDC access key management functions. (#857)
*   **MinIO Client Stability**: Resolved a crash in the MinIO Client (mc) that occurred when displaying certain diagnostic data in a table format. (#864)

### Security Updates

*   **Hardened Kubernetes Authentication**: Enhanced security by defaulting the expected audience for Kubernetes service account tokens to `sts.min.io`. A new configuration option is available to manage the token audience for backward compatibility. (#674)