At the end of November, we’ll be migrating the Sematext Logs backend from Elasticsearch to OpenSearch

Introducing Kubernetes monitoring – new reports for all K8S components!

August 2, 2023

We are excited to announce the latest release of Sematext Kubernetes Monitoring. This much improved Kubernetes monitoring integration brings a comprehensive suite of dashboards for some of the most critical components of your Kubernetes infrastructure, including the Control Plane, Node, and Addons.

To start leveraging these new reports and optimizing your Kubernetes environment, simply update to the latest version of Sematext Agent. Read about all the new goodies below or check out the Kubernetes monitoring docs.

API Server Reports

Monitor health and performance of Kubernetes API Server with four new out-of-the-box reports displaying more than 75 new metrics, including:

  • Request Latency: Measures the time taken to process API server requests
  • Request Throughput: Tracks the number of API server requests processed per unit of time
  • Error Rate: Monitors the rate of API server errors

See all API Server metrics collected.

Schedulers Reports

Monitor the behavior of your Kubernetes schedulers, identify scheduling-related issues, and ensure your workloads maintain a healthy, balanced cluster with five out-of-the-box reports featuring more than 40 metrics, including:

  • Scheduling Latency by Attempts: Measures the scheduling latency for pods based on the number of attempts made
  • Failed Scheduling Attempts: Monitors the number of failed pod scheduling attempts
  • Queued Pods by Queue: Tracks the number of pods currently in the scheduler’s queue, categorized by the queue name
  • unschedulable Pods: Tracks the number of pods that cannot be scheduled due to resource constraints

See all Scheduler metrics collected.

CoreDNS Reports

Obtain visibility into DNS query performance, response times, cache hits/misses, and overall DNS health. Detect potential DNS-related issues and ensure uninterrupted application performance with four pre-configured dashboards offering over 25 metrics, such as:

  • DNS Request Latency: Measures the time taken to process DNS requests by CoreDNS
  • DNS Local and Remote Cache Misses: Counts the number of cache misses for DNS queries in CoreDNS’s local or remote cache
  • Error Rate: Monitors the rate of DNS errors encountered by CoreDNS

See all CoreDNS metrics collected.

kube-proxy Reports

The six new kube-proxy reports give you detailed insights into how kube-proxy is handling network traffic, helping you detect any networking issues early on and optimize service-to-service communication. Some example kube-proxy metrics include:

  • Service Changes: Tracks the number of changes in services detected by kube-proxy
  • Endpoint Changes: Tracks the number of changes in endpoints detected by kube-proxy
  • Synchronization of Proxy Rules: Measures the time taken to synchronize proxy rules for services
  • Request Latency by Host, HTTP Method, Path: Measures the latency of requests proxied by kube-proxy, categorized by host, HTTP method or Path

See all kube-proxy metrics collected.

Etcd Reports

Etcd is the backbone of your Kubernetes control plane, storing all critical cluster data. The five new Etcd reports provide essential visibility into Etcd’s performance, resource consumption, and cluster state, with more than 60 metrics, including:

  • Leader Changes: Tracks the number of times the etcd cluster leadership changes
  • Disk Space Usage: Monitors the amount of disk space used by etcd
  • WAL Write Latency: Measures the latency of write operations to the etcd Write-Ahead Log (WAL)
  • WAL Snapshot Latency: Measures the latency of taking snapshots of the etcd Write-Ahead Log (WAL)
  • sss
  • WAL Commit Latency: Measures the latency of committing changes from the etcd Write-Ahead Log (WAL) to the database

See all Etcd metrics collected.

Kubelet Reports

With four new kubelet reports and over 100 metrics, you can monitor node health and performance and gain quick insights into pod and container start and runtime errors with a dedicated Errors report. Other notable metrics include:

  • Kubelet Instances: Count of kubelet agents running on individual nodes
  • Running Pods and Containers: These metrics provide the current count of running pods and containers on the node
  • Pod Lifecycle Event Generator (PLEG): Mechanism that monitors and manages the lifecycle of pods
  • Container Runtime Operations: the total number of container runtime operations performed by the kubelet

See all kubelet metrics collected.

For more information, please refer to Kubernetes Monitoring Integration, and don’t forget to update to the latest version of Sematext Agent to get all these new reports.