Use OCP-Agent to pull time-series monitoring data

OCP-Agent can push the monitoring data to Pushgateway, which can pull monitoring data collected by OCP-Agent.

OCP-Agent monitoring description

OCP-Agent is a management and monitoring component deployed on the host. It runs with three resident processes and one command-line tool (CLI) process.

• The ocp_mgragent process is the management process that executes O&M commands for the OceanBase database and OBProxy.

• The ocp_monagent process is the monitoring process that collects monitoring data from the OceanBase database, OBProxy, and host.

• The ocp_agentd process is the daemon process of ocp_mgragent and ocp_monagent.

• The ocp_agentctl process is the process of the O&M CLI tool.

OCP-Agent generates the following two types of time-series monitoring data in the formats defined in Prometheus:

1. Business monitoring data, such as the monitoring data of hosts, OceanBase database, clusters, and tenants.

2. OCP-Agent self-monitoring data related to the running status of OCP-Agent, such as the resource usage by OCP-Agent and the running status of the related processes.

Scheme for pulling time-series monitoring data

In OCP V3.2.0 and later, the monitoring APIs of OCP-Agent support authentication. The authentication uses random passwords that cannot be queried. In OCP V3.3.2, you can disable the authentication. After the authentication is disabled, OCP-Agent can directly pull the monitoring data.

The monitoring APIs are controlled by the ocp.agent.auth.metric-auth-enabled parameter. The default value is true, which indicates that the authentication is enabled. If you set the parameter to false, OCP-Agent can directly pull the monitoring data.

The following table describes the monitoring APIs: