@Rocky what grafana dashboard templates do you have and whats are the differences?

StarRocks provides several Grafana dashboard templates tailored to different cluster architectures and monitoring needs. You can generally find the download links for these JSON templates in the Monitor and Alert section of the documentation. The primary differences between them are: *1. All Architecture (General) Template* * Best for: Shared-nothing clusters (standard deployment) and basic monitoring for any StarRocks version. * Focus: Provides a high-level overview of the cluster. It includes core metrics for both FE (Frontend) and BE (Backend) nodes, such as: * Node status (alive/dead). * Query latency and throughput (QPS). * CPU, Memory, and Disk usage. * JVM statistics for FE. * Basic loading (ETL) metrics. *2. Shared-data Cluster Templates* Shared-data clusters (disaggregated storage/compute) use specialized dashboards because their architecture differs significantly from the standard version. * Shared-data General Dashboard: Focuses on the health and performance of the Compute Nodes (CN) and the overall shared-data cluster state. * Starlet Dashboard: This is a specialized dashboard for the "Starlet" component (the storage abstraction layer in shared-data mode). It focuses heavily on storage I/O and object storage interactions: * I/O Metrics: S3/Object storage read/write latency, throughput, and IOPS. * Data Cache: Cache hit/miss ratios and cache disk/memory usage. * FileSystem (FSLIB): Metadata operations and asynchronous task queues. *3. Environment-Specific Templates (Shared-data)* For users running shared-data clusters in specific environments, StarRocks provides optimized templates: * Shared-data for VM: Tailored for virtual machine deployments. * Shared-data for Kubernetes (K8s): Optimized for containerized environments, often including metrics related to pod resource constraints. Key Considerations * Detailed Metrics: By default, dashboards show cluster-level metrics. If you need table-level, Materialized View, or user-specific connection metrics, you must enable them in your Prometheus scrape configuration using parameters like

with_table_metrics=all

and provide ADMIN credentials. * Version Compatibility: Always download the template version that matches your StarRocks version (e.g., 3.x vs 2.x) to ensure all metric names (PromQL) are compatible. References • administration/management/monitoring/metrics-shared-data.md • deployment/shared_data/feature-support-shared-data.md • deployment/shared_data/azure.md