Cluster Inspection #
With the increasing complexity of Kubernetes system and its deployment,the existing observation system will encounter the following problems:
- As a forward link, monitoring data can’t cover all scenes without dead ends.
- Even if the monitoring data of each node in the link is normal, the link can’t be 100% available.
- Consistency of data flattening in kubernetes cluster.
In the past,in order to solve the above-mentioned similar problems,such as OOM detection, kubelet anomaly detection, etcd object detection, etc.,but these empirical things like systems are scattered everywhere.In order to ensure the cluster stability in complex scenes and form an “expert experience base”,we plan to build Kubernetes cluster Synthetic Monitoring system,which simulates the behavior of generalized “users” from the angle of black box to actively detect clusters/components/link objects,and supports directional inspection.
Operating steps:
Log in the MSP Platform
In the MSP Platform “All Products” navigation bar,click the Telemetry under the Container Cloud Platform module,click Cluster Inspection in the right navigation bar,as shown in the following figure:
