Start Monitoring Center
The Monitoring Center supports both single node mode and high availability mode. It should be noted that enabling monitoring consumes a certain amount of CPU and memory resources. Therefore, if you have enabled the high availability mode, please note:
-
At least 2 Node nodes have available resources greater than the container configuration of Prometheus. (It is recommended that available resources be greater than 4C8G)
-
At least 3 Node nodes have available resources greater than the container configuration of Alertmanager. (It is recommended that available resources be greater than 1C2G)
-
As both Prometheus and Alertmanager require persistent storage, this will generate additional UDisk costs. Prometheus for 2 100G UDisks and Alertmanager for 3 UDisks.
Turn on monitoring
The recommended parameter configuration is as follows:
- Prometheus data disk size: more than 100G. If the cluster size exceeds 100, it is recommended to expand the disk size to 500G;
- Prometheus data retention duration: it is recommended to be 240 hours;
- Grafana configuration: Both the username and password can be customized;
External access to monitoring
When enabling external access, you can create LB resources by selecting the LB type or use existing resources to call the API to enable external access. By default, CLB7 will be used. When using existing resources, ensure that the ports do not conflict (turning on external access to monitoring will occupy ports 80, 9090, 9093). CLB7 can be used for free, but with certain restrictions, it is recommended to use ALB. More details can be found in the CLB documentation and ALB documentation.