Download PDF
Download page Clusters.
Clusters
The Clusters page acts as a status page for real-time and historical data on the performance of Kubernetes® clusters, including the following:
- Metrics at the cluster level
- Summary of metrics from the namespaces, workloads, PVCs, pods, and containers
- Summary of changes happened in Kubernetes Configurations
- Kubernetes Events
- Network connections, if you have enabled Network Monitoring.
For information on the network monitoring user interface and metrics, see Monitor Network Connections for Kubernetes Workloads.
Cisco Cloud Observability monitors the health status, attributes, metrics, and relationships of each cluster. It provides metrics that the Kubernetes Infrastructure Collector derives from examining your backend targets.
You connect the Cisco Cloud Observability Kubernetes Collectors to Cisco Cloud Observability. See Install Kubernetes and App Service Monitoring.
You can use Clusters to:
- Determine if there are any issues relating to overall cluster capacity and resource usage.
- Gain insight into application issues that affect services running inside the cluster.
- Diagnose issues that are affecting the scalability of services.
- Determine if there are any changes in Kubernetes Configurations (ConfigMaps and Secrets)
This document contains references to the Amazon Web Services (AWS) documentation. AppDynamics does not own any rights and assumes no responsibility for the accuracy or completeness of such third-party documentation.
If the cluster you connect to Cisco Cloud Observability already has Splunk AppDynamics APM agents running on it, the metrics you see on the Observe page may be different from the metrics you see on your Splunk AppDynamics APM Controller. Comparing metrics on the two platforms may yield inconclusive results. There are multiple reasons for this. Cisco Cloud Observability only ingests metrics from AWS CloudWatch, Azure Monitor, and OpenTelemetry™-compatible collectors, whereas Splunk AppDynamics App Server agents, such as Machine Agent, send metrics in a proprietary, non-OpenTelemetry format. In addition, those agents are connected to your Controller UI, not to Cisco Cloud Observability.
Metrics and Key Performance Indicators
Cisco Cloud Observability displays the following metrics and key performance indicators (KPIs) for clusters list and detail views:
| Display Name | Description |
|---|---|
Pod Status Events | The status of the pods reported every minute. |
Memory Usage vs Limit (%) | The total memory used versus memory limits in a cluster. Here, the total memory used is the memory used by the pods that include the memory limit. |
Memory Usage vs Request (%) | The total memory used versus memory requests in a cluster. Here, the total memory used is the memory used by the pods that are configured with the memory request. |
CPU Usage vs Limit (%) | The total CPU used versus CPU limits in a cluster. Here, the total CPU used is the CPU used by the pods that are configured with the CPU limit. |
CPU Usage vs Request (%) | The total CPU used versus memory requests in a cluster. Here, the total CPU used is the CPU used by the pods that are configured with the CPU request. |
Host Memory Usage (%) | The percentage of memory used of allocated host memory. |
Host CPU Usage (%) | The percentage of CPU used of allocated host CPU. |
PV Usage (GiB) | The total persistent volume (in GiB) used in a cluster. Click Show PVCs to view the breakdown of each PV Usage vs Request by PVC at cluster level. |
ConfigMaps and Secrets Changelog
Cisco Cloud Observability displays any change that happened in Configurations (ConfigMaps and Secrets) in real time.
Properties (Attributes)
Clicking a cluster Name displays the detail view and the Properties panel with these properties (attributes) for clusters:
| Display Name | Description |
|---|---|
| Name | The cluster name. |
| Created At | The time when cluster was created. |
| Readiness Probes | The number of pods with readiness probes configured against the total number of pods. |
| Liveness Probes | The number of pods with liveness probes configured against the total number of pods. |
| K8s Version | The Kubernetes version. |
The Properties panel shows limited metadata.
You can use Filter View to filter using any of the available k8s attributes.
You can use the k8s.cluster.id attribute in the Filter View field to observe entity details for a cluster that does not have a unique name. In the filter, specify the value of this attribute as the UUID of the kube-system namespace. For example, attributes(k8s.cluster.id) = <uuid of kube-system namespace>.
The k8s.cluster.id attribute is not displayed in the Properties panel, but you can use the key value pair to filter the data based on a unique cluster ID.
Tags
Tags are labels consisting of key-value pairs. Some Kubernetes attributes are promoted to tags and these tags are propagated to other entities. See Tags.
Clicking a cluster Name displays the detail view and the Tags panel that lists propagated tags, Kubernetes labels, and any imported tags that you set up during cloud connection configuration.
You can filter entities based on tags.
If you experience the following issues during Kubernetes Cluster monitoring, increase the hop limit to two by following the instructions mentioned in the AWS User Guide:
- The hostname is incorrect
- The AWS host is appearing in the Onprem tab on the Host list page
Retention and Purge Time-to-Live (TTL)
See Retention and Purge Time-to-Live (TTL).
OpenTelemetry™ and Kubernetes® (as applicable) are trademarks of The Linux Foundation®.