Grafana Monitoring

YMatrix provides a native graphical monitoring tool based on Grafana. This document describes the steps to deploy and manage the monitoring components. YMatrix includes a default monitoring dashboard that displays the following information. Users can also create custom dashboards in Grafana using the collected system data.

  • Cluster status, version, current connection count, uptime, and data node (Segment) status
  • Disk space: current disk usage on the master node (Master) and Segment hosts
  • Database logs: view recent database logs, filterable by severity levels such as warning, error, critical, and fatal
  • Recent system load on database servers, including:
    • CPU, memory, disk I/O, network I/O, and process count
    • Ability to select all or specific hosts for monitoring
    • Flexible time range selection and auto-refresh intervals

After deployment, the monitoring interface appears as shown below. Starting from version 4.4, the monitoring dashboard is split into two panels: MatrixDB Dashboard and MatrixDB Database.

MatrixDB Dashboard:
MatrixDB Dashboard

MatrixDB Database:
MatrixDB Database

1 Deployment

Monitoring components are included in the YMatrix installation package. After deploying YMatrix, complete the monitoring setup in two steps: enabling metric collection and installing and configuring Grafana.

1.1 Enable Metric Collection

Perform the following steps to enable metric collection for YMatrix and system resource usage. Collected data is stored in a newly created database named matrixmgr. In the examples below, the master node is named mdw.

  • Switch to the mxadmin user
[<普通用户>@mdw ~]$ sudo su - mxadmin
  • Connect to the matrixmgr database and enable metric collection
[mxadmin@mdw ~]$ psql -d matrixmgr
matrixmgr=# SELECT mxmgr_init_local();

Upon success, a new schema named local appears in the matrixmgr database. Tables and views under this schema contain cluster monitoring and configuration data. Do not modify the definitions or contents of these tables and views manually.

1.2 Install and Configure Grafana

Prepare a host that can access both the Master node and the internet. This can be the Master, Standby Master, or a separate machine (Linux, macOS, Windows, etc.).

Install Grafana, version 7.3 or later is recommended. The official download link is https://grafana.com/grafana/download.

Commands below use CentOS 7 as an example. For other operating systems, refer to the corresponding documentation.

Note!
YMatrix supports offline installation of Grafana. For details, see Monitoring and Alerting FAQ.

  • Download and install Grafana
wget https://dl.grafana.com/oss/release/grafana-7.3.6-1.x86_64.rpm
sudo yum install grafana-7.3.6-1.x86_64.rpm
  • Start Grafana
[mxadmin@mdw ~]$ sudo systemctl daemon-reload
[mxadmin@mdw ~]$ sudo systemctl start grafana-server
[mxadmin@mdw ~]$ sudo systemctl status grafana-server
[mxadmin@mdw ~]$ sudo systemctl enable grafana-server

Note!

  1. The Grafana version provided by yum in CentOS 7 is often outdated (6.x), so sudo yum install grafana is not recommended.
  2. For the complete official installation guide, see https://grafana.com/docs/grafana/latest/installation/rpm

After installation, access the following URL in a browser. Port 3000 is the default Grafana port and can be changed. Log in using the default credentials (admin / admin). For security, change the password after login by clicking the user icon in the lower-left corner.

http://<IP_or_domain_of_installed_node>:3000
  • Configure monitoring dashboards

After Grafana is installed, add the matrixmgr database in YMatrix as a data source and import the predefined monitoring dashboards.

Steps to add a data source:
Add Data Source 1
Add Data Source 2
Add Data Source 3

Before importing the predefined dashboards, copy the dashboard.json and database.json files to your local machine. The process is: locate the files on the server, copy them, and then upload them locally. Follow these steps:

First, log in to the server and switch to the mxadmin user. Locate the dashboard.json and database.json files in the path shown below. You can use the cd command, find command, or any other method you prefer. The find command example below searches for database.json.

[mxadmin@mdw ~]$ cd /opt/ymatrix/matrixdb5/share/en/doc/postgresql/extension

##or

[mxadmin@mdw ~]$ find /opt/ymatrix/matrixdb5/share/en/doc/postgresql/extension -name database.json

Next, use the scp command to copy the files to your local machine. Permission issues may arise. Consider copying the file to the /tmp/ directory first, then from /tmp/ to your local machine. The example below uses database.json.

Note!
When copying from /tmp/, ensure you switch users to avoid permission issues.

[mxadmin@mdw]$ scp mxadmin@<server_IP>:/opt/ymatrix/matrixdb5/share/en/doc/postgresql/extension/"database.json" mxadmin@<server_IP>:/tmp/

~ scp waq@<server_IP>:/tmp/"database.json" /Users/akkepler/workplace/Grafana

Finally, open the local folder or use the command line to verify the file was copied successfully. After confirmation, upload the files in the Grafana interface.

Note!
If the import fails, perform the following variable modifications.

In dashboard.json, replace all instances of ${cluster} with local. Replace $host with the actual hostnames. For example, if your cluster consists of mdw, sdw1, and sdw2, change all $host entries in dashboard.json to 'mdw','sdw1','sdw2'. For database.json, only replace ${cluster} with local. After modification, import the files in Grafana.

Import Dashboard 1

Click Upload JSON file to select and import the dashboard.json and database.json files. These files are located by default in /opt/ymatrix/matrixdb5/share/en/doc/postgresql/extension.

Import Dashboard 3

2 Management

After enabling cluster metric collection, each host runs a collection service. Related logs are stored in the /var/log/matrixdb directory.

If you restart YMatrix or reboot the server and then restart YMatrix, the data collection service starts automatically. No manual intervention is required.

To stop the data collection service, connect to the matrixmgr database and run mxmgr_remove_all. Collected data remains preserved after stopping:

[mxadmin@mdw ~]$ psql -d matrixmgr

matrixmgr=# SELECT mxmgr_remove_all('local');

If the collection service is stopped manually or after reinstalling YMatrix, re-enable collection by connecting to the matrixmgr database and running mxmgr_deploy:

matrixmgr=# SELECT mxmgr_deploy('local');