Robusta KRR is a command-line tool designed to optimize resource allocation in Kubernetes clusters by analyzing pod usage data from Prometheus and recommending appropriate CPU and memory requests and limits. The tool aims to reduce cloud infrastructure costs and improve application performance through intelligent right-sizing of container resources. According to a Sysdig study referenced in the repository, Kubernetes clusters typically have 69% unused CPU and 18% unused memory on average, making KRR's optimization recommendations potentially valuable for significant cost savings.
The tool distinguishes itself from the native Kubernetes Vertical Pod Autoscaler through several key differences. Unlike VPA, KRR does not require installation within the cluster and can run locally on a developer's machine while connected to a cluster. It provides immediate results without requiring time to gather baseline data, and it does not necessitate creating VPA objects for each workload. KRR offers multiple reporting formats including JSON, CSV, Markdown, and a web UI, along with explainability features that show graphs explaining how recommendations were calculated. The tool supports HPA-enabled workloads through an allow-hpa flag and maintains a default 14-day history compared to VPA's 8-day window.
KRR integrates with multiple Prometheus-compatible data sources including Thanos, Victoria Metrics, Google Managed Prometheus, Amazon Managed Prometheus, Azure Managed Prometheus, Coralogix, Grafana Cloud, and Grafana Mimir. The tool requires Prometheus 2.26 or later along with kube-state-metrics and cAdvisor, though no additional setup is needed if using kube-prometheus-stack. It relies on specific metrics including container_cpu_usage_seconds_total, container_memory_working_set_bytes, kube_replicaset_owner, kube_pod_owner, and kube_pod_status_phase.
Installation options include Homebrew for Mac and Linux, Docker containers, pre-built binaries for offline environments, in-cluster deployment via the Robusta Platform, and installation from source using Python 3.9 or greater. The repository includes an auto-apply mode through KRR Enforcer that can automatically apply recommendations to put right-sizing on autopilot. Reporting integrations extend functionality to Slack notifications, k9s plugin integration, and Azure Blob Storage export with Teams notifications.
According to GitGenius activity tracking, the repository shows median issue and pull request response latency of 22.3 hours across 108 tracked items, with a mean latency of 1850.9 hours indicating some longer-running discussions. The most active contributor is aantn with 190 tracked events, followed by arikalon1 with 28 events and patsevanton with 17 events. Enhancement requests represent the most common issue label with 3 instances, followed by bug reports with 2 instances and good first issue designations with 1 instance. The repository's contributor network overlaps with major projects including Microsoft's VSCode and TypeScript repositories as well as the Rust language repository, suggesting involvement from developers across significant open-source ecosystems.
