Prometheus monitoring|V4.3.2|OceanBase Database Proxy |docs|OceanBase Documentations

In addition to a log system, OceanBase Database Proxy (ODP) also provides monitoring data for integration with different monitoring platforms through Prometheus.

You can configure the following parameters in ODP to manage and adjust the Prometheus monitoring feature.

Parameter	Description
prometheus_listen_port	The Prometheus listening port of ODP. Default value: `2884`.
prometheus_cost_ms_unit	Specifies whether to convert the calculation unit of Prometheus to milliseconds. The default unit is microseconds. If you set this parameter to `true`, the time unit of monitoring data changes to milliseconds.
prometheus_sync_interval	The time interval for refreshing Prometheus monitoring data. Default value: `1 min`.
enable_prometheus	Specifies whether to enable Prometheus monitoring. Default value: `true`.
enable_extra_prometheus_metric	Specifies whether to enable other metrics. Default value: `false`. Recommended value for a test environment: `true`.

After you correctly configure related Prometheus parameters, namely, retain the default values for the parameters except enable_extra_prometheus_metric, start ODP, connect to the OceanBase cluster through ODP, and execute several SQL statements. You can run the curl "http://127.1:2884/metrics" command to view related monitoring data of Prometheus.

Basic metrics

Metric	Description	Tag	Remarks
odp_transaction_total	The number of transactions that have been executed. The value monotonically increments.	cluster, tenant, schema	None
odp_sql_request_total	The number of SQL statements that have been executed. The value monotonically increments.	cluster, tenant, schema, slowQuery, sqlResult, sqlType, partitionHit	`sqlResult` indicates the execution result and has two valid values: "fail" and "success". ODP provides a new tag `partitionHit` in V4.3.0 and later to indicate whether routing is accurate. The tag has two valid values: "true" and "false". In ODP V4.3.0 and later, the `odp_sql_request_total` metric supports the `sqlType` tag in the public cloud. This tag has the following valid values: select, update, insert, delete, and other.
odp_current_session	The current number of connections of ODP, including client connections to ODP and ODP connections to the OBServer node. The value does not monotonically increment.	cluster, tenant, schema, sessionType, vip	`sessionType` has two valid values: "client" and "server", which respectively record the number of connections between the client and ODP and that between ODP and the OBServer node. ODP provides a new tag `vip` in V4.3.0 and later to indicate the client VIP. Note This metric does not record the number of connections of the `proxyro@sys` user.
odp_new_client_connections	The number of client connections with login attempts to ODP. The value monotonically increments.	cluster, tenant, vip, ConnectionResult	This metric was introduced in ODP V4.3.0. ODP provides a new tag `ConnectionResult` in V4.3.0 BP1 and later to indicate whether the login is successful. The tag has two valid values: "fail" and "success". Note This metric does not record the number of connections of the `proxyro@sys` user.
odp_sql_cost_total	The time consumed to execute the current SQL query. The value monotonically increments.	cluster, tenant, schema, timeType, sqlType	`timeType` has three valid values: "prepare": the time from when ODP receives the SQL query to when the query is forwarded to the OBServer node. "server": the processing time of the SQL query on the OBServer node. "total": the time from when ODP receives the SQL query to when all response packets are forwarded to the client. In ODP V4.3.0 BP1 and later, the `odp_sql_cost_total` metric supports the `sqlType` tag in a public cloud. This tag has the following valid values: select, update, insert, delete, and other.
odp_memory	The memory information about ODP.	memType	`memType` has two valid values: HOLD: the memory requested by ODP from the operating system, which is comprised of two parts: the actually used memory and the cache memory. USED: the memory actually used by ODP.

Extra metrics

By default, the metrics listed in the following table do not have data. They provide data only when enable_extra_prometheus_metric is set to true.

Metric	Description	Tag	Remarks
odp_request_byte	The number of bytes transmitted between the client and ODP and between ODP and the OBServer node. The value monotonically increments.	cluster, tenant, schema, sessionType, transType	`sessionType` has two valid values: "client" and "server", which respectively record the number of bytes transmitted between the client and ODP and that between ODP and the OBServer node. `transType` has two valid values: "request" and "response", which respectively indicate request data and response data. `sessionType` can be used in combination with `transType` to represent four types of byte flows. Note This metric exists only in a private cloud environment.
odp_entry_total	The number of times that ODP searches for `table_entry` and `partition_entry`.	cluster, tenant, schema, entryType, routeHit, routeResult	`entryType` has three valid values: "table_entry", "partition_info", and "partition_entry". `routeHit` has two valid values: "true" and "false". The value "true" indicates that the information is obtained from the local cache. `routeResult` has two valid values: "true" and "false". The value "true" indicates that the information is successfully obtained.

Related parameters

Basic metrics

Note

Note

Extra metrics

Note