This topic describes the metrics of E-MapReduce (EMR) Serverless StarRocks monitored by CloudMonitor. The metrics involve multiple key performance parameters.
Availability
Metric name | Type | Dimension | Description |
| Instantaneous value | - | The liveness rate (Alive) of backend (BE)/compute node (CN) nodes. This metric reflects the health status of the nodes. |
| Instantaneous value | podName | Indicates whether the startup log of a frontend (FE) process is captured. A value of 1 means that the characteristic log of a process startup is detected during the time period. |
| Instantaneous value | podName | Indicates whether the startup log of a BE/CN process is captured. A value of 1 means that the characteristic log of a process startup is detected during the time period. |
| Instantaneous value | podName | The result of an HTTP request to the node. A value of 1 indicates success. A value of 0 indicates failure. This metric can be used to indicate the status of an FE node. |
| Instantaneous value | podName | The result of an HTTP request to the node. A value of 1 indicates success. A value of 0 indicates failure. This metric can be used to indicate the status of a BE/CN node. |
Resource usage
Metric name | Type | Dimension | Description |
| Instantaneous value | podName | The CPU utilization of the FE node. |
| Instantaneous value | warehouseName podName | The CPU utilization of the BE node. |
| Instantaneous value | podName | The Java Virtual Machine (JVM) heap memory usage of the FE node. |
| Instantaneous value | podName | The memory usage of the FE node. Due to the characteristics of the JVM, use the |
| Instantaneous value | warehouseName podName | The memory usage of the BE node. |
| Instantaneous value | warehouseName podName | The size of memory allocated to the BE node. Unit: Byte. |
| Instantaneous value | warehouseName podName | The total memory used during query execution. |
| Instantaneous value | warehouseName podName | The total memory used by import tasks. |
| Instantaneous value | warehouseName podName | The memory used by version compaction. |
| Instantaneous value | warehouseName podName | The memory used by Schema Change tasks. |
| Instantaneous value | warehouseName podName | The column pool memory pool. This is a Column Cache used to accelerate data reads from the storage layer. |
| Instantaneous value | warehouseName podName | The page cache of the BE storage layer. |
| Instantaneous value | warehouseName podName | The cache for BE just-in-time (JIT) compiled functions. |
| Instantaneous value | warehouseName podName | The per-core CPU cache. This cache is used to accelerate small memory block requests. |
| Instantaneous value | warehouseName podName | The memory used for periodic consistency checks. |
| Instantaneous value | warehouseName podName | The memory used by cached data. |
| Instantaneous value | warehouseName podName | The total memory used by Tablet Clone tasks. |
| Instantaneous value | warehouseName podName | The memory used by metadata. |
| Instantaneous value | podName | The disk capacity usage of the BE node. |
| Instantaneous value | warehouseName podName | The maximum disk I/O utilization. |
| Instantaneous value | podName | The total number of connections to the FE node. |
| Cumulative value | warehouseName podName device | The total amount of data sent over the network. Unit: Byte. |
| Cumulative value | warehouseName podName device | The total amount of data received over the network. Unit: Byte. |
Query performance
Metric name | Type | Dimension | Description |
| Instantaneous value | podName | The percentile statistics of query latency, such as P75, P95, and P99. Differentiated by the |
| Instantaneous value | podName name type | The latency statistics for resource group queries. |
| Instantaneous value | podName | The number of queries executed per second. |
| Instantaneous value | podName | The number of failed queries per minute. |
| Instantaneous value | podName | The number of slow queries per minute. |
| Cumulative value | podName | The total number of query requests. |
| Cumulative value | podName | The total number of failed requests. |
| Cumulative value | podName | The total number of timed-out queries. |
| Instantaneous value | podName | The number of queries currently in the queue. |
| Instantaneous value | podName | The number of historical queued queries, including queries that are running. |
| Instantaneous value | podName | The total number of queries that timed out in the queue. |
Transaction management
Metric name | Type | Dimension | Description |
| Instantaneous value | db | The number of currently running transactions. |
Materialized view (MV)
Metric name | Type | Dimension | Description |
| Instantaneous value | dbName mvId mvName | The status of the materialized view. Valid values: |
| Instantaneous value | dbName mvId mvName | The number of refresh jobs that are currently running for the materialized view. |
| Instantaneous value | dbName mvName | The number of failed materialized view (MV) refresh jobs per minute. |
| Instantaneous value | dbName mvName | The number of failed MV refresh jobs. |
| Cumulative value | dbName mvId mvName | The number of pending refresh jobs for the MV. |
Other key metrics
Metric name | Type | Dimension | Description |
| Instantaneous value | podName exportedJob type state | The number of current jobs. |
| Instantaneous value | podName | The number of Berkeley DB (BDB) metadata logs for the FE. |
| Instantaneous value | podName | The maximum number of versions for a tablet in the BE. A high value may affect performance. |
| Instantaneous value | podName type | The duration of a slow DB lock. |
| Instantaneous value | warehouseName podName | The 99th percentile (P99) latency of Publish Version tasks. |
| Instantaneous value | - | The highest Compaction Score on a BE node. This reflects the compaction pressure. |