Learn about Kubernetes monitoring using Prometheus and Grafana
- Understand the why aspect of monitoring and the advantages it offers
- Install Prometheus and Grafana on a Kubernetes cluster
- Monitor a Mini Cube Kubernetes cluster using Prometheus and visualize the data with Grafana
Monitoring Kubernetes clusters using Prometheus and Grafana
- Prepare a Grafana dashboard to show metrics of API servers and deployments
- Understanding the need for monitoring with multiple Kubernetes clusters
Prometheus stores metrics in a time series database and has a built-in monitoring system.
- Prometheus stores metrics of Kubernetes cluster resources based on timestamps.
- Prometheus can be configured with alert manager to send notifications to different platforms.
Prometheus and Grafana provide monitoring and visualization for Kubernetes
- Prometheus offers a UI to execute Prometheus queries and get recorded information
- Grafana allows easy visualization of the data retrieved from Prometheus
Installing Prometheus using Helm
- Ensure Kubernetes cluster is set up
- Copy and paste the commands from the GitHub page to add Prometheus Helm chart
Install Prometheus controller and required configurations using Helm
- Ensure to update Helm repository before installation
- Read important information provided during installation for server URL and other details
Exposing the Prometheus server UI
- Convert the cluster IP mode service to a node Port service using the provided command
- Access the Prometheus server UI using the node port 31110 on the Kubernetes node
Prometheus can be installed on a Kubernetes cluster to collect and query metrics.
- Prometheus can capture only the metrics exposed by the Kubernetes API server by default.
- To get more detailed information about an application, developers can write a metric server or use Prometheus Matrix libraries to expose a metrics endpoint.
Exposing Grafana using Node Port mode
- To expose Grafana and access it externally, Cube CTL is used to create a Node Port mode service.
- In production or staging environments, Ingress or operator can be used to create routes or automatically create the service.
Setting up Prometheus and Grafana for Kubernetes monitoring
- Create Prometheus as a data source in Grafana
- Configure the IP address of Prometheus and save and test the configuration
Standard template (ID: 3662) in Grafana provides information about Kubernetes cluster
- Template fetches data from Kubernetes API server and nodes
- Queries in template include uptime of miniCube cluster, memory chunks, and misdictions
Using Cube State Matrix to get detailed information about the Kubernetes service
- Cube State metrics provides information about deployment status, running replicas, and current status of the Kubernetes service
- To expose the Cube State Matrix, use the command 'expose' followed by the target port which is 8080
- Access the Cube State Matrix using the MiniKube IP address and port 30421
Expose the cube State metrics and configure Prometheus to scrape data from a specific endpoint
- To expose the cube State metrics, use the endpoint 192.168.64.15:<nodePort>/metrics
- To configure Prometheus to scrape data from this endpoint, edit the 'Prometheus server' config map and add a new job with the target IP address
Monitoring applications in Kubernetes using Prometheus and Grafana
- Developers can write a matrix server to get the health information of their applications
- Prometheus and Grafana can be used for monitoring and visualization
Launch your Graphy