klusterview

Klusterview

Klusterview scans Kubernetes cluster to visualize resources & utilization

Compiling

go build .

Running

• First have Redis with Time Series module enabled, e.g: docker run -d --name redis -p 6379:6379 redis/redis-stack

• Run a sample: ./klusterview This will populate Redis with the sample data.

• Run Webdis to expose Redis on HTTP port (7379).

• Run simple web server on localhost (e.g: php -S 0.0.0.0:8000) and browse there.

Data Model

Hybrid data model - Basic Redis data types + Time Series keys

Keys:

• samples - total samples count in db (int)
• sample:i - timestamp of beginning of cluster scan (int)
• sample:i:nodes - node names existing at time of scan (set)
• sample:i:pods - pod names existing at time of scan (set)
• node:node_name:alloc_cpu - allocatable cpu for specific node (int)
• node:node_name:alloc_mem - allocatable memory for specific node (int)
• node:node_name:util_cpu - utilization of cpu (Time Series)
• node:node_name:util_mem - utilization of memory (Time Series)
• pod:pod_name - pod meta data (dictionary)
• pod:pod_name:containers - containers (set)
• pod:pod_name:container_name:req_cpu - requested cpu (int)
• pod:pod_name:container_name:req_mem - requested memory (int)
• pod:pod_name:container_name:util_cpu - utilization of cpu (Time Series)
• pod:pod_name:container_name:util_mem - utilization of memory (Time Series)

Pipeline Stages

• Scan - Scan cluster Nodes & Pods to extract resource requests & utilization
• Ingest - Write data to Redis