This is the branch where we publish Helm charts to GitHub Pages.
OOM Tracer observes global Linux OOM killer events on Kubernetes nodes, maps them back to pods, and annotates the offending workloads so that a descheduler instance can evict them. The intent is to reduce cascading failures that follow severe memory pressure by letting you plug the detection signal into an AnnotationEvictor downstream.
To install OOM Tracer using Helm, add the Helm repository:
helm repo add oom-tracer https://eminaktas.github.io/oom-tracer/
helm repo update
helm install oom-tracer oom-tracer/oom-tracer --namespace oom-tracer-system --create-namespace
For complete documentation on OOM Tracer, including configuration options, examples, and advanced usage, refer to the OOM Tracer README.
We welcome contributions to OOM Tracer! If you’d like to contribute, please fork the repository and submit a pull request.
This project is licensed under the MIT License.