BeaverDeck logo
BeaverDeck

Home / BeaverDeck / Docs / Insights Guide / Workload Insights / OOMKilled

OOMKilled

BeaverDeck uses this check to identify a specific workload condition that may need operator review.

Permissions: viewing checks requires insights: view. Opening a linked object or logs requires the corresponding resource permission, and the BeaverDeck ServiceAccount must be allowed to read the Kubernetes resources used by the check. Suppressing a finding requires insights: edit and affects all users.
Check typeoomkilled
Insights sectionWorkload Insights
Alert severityWarning

When It Reports A Finding

The current or previous terminated state of an init or application container reports reason OOMKilled.

Why This Is A Problem

The process exceeded available memory and was forcibly terminated. Repeated kills cause downtime and can expose leaks, unsafe limits, or unexpected workload growth.

Recommended Response

  1. Inspect the container's memory request, limit, restart history, and memory usage before termination.
  2. Determine whether the cause is a leak, an expected peak, unbounded caching, or an undersized limit.
  3. Fix application memory behavior where possible; otherwise adjust requests and limits using measured demand and node capacity.

Scope And Limitations

The finding records an OOM termination but cannot determine whether the correct fix is application work, a higher limit, or workload scaling. Raising limits without investigation can move pressure to the node.

After remediation: refresh Workload Insights and verify the underlying resource or metric. Suppress the finding only when the condition is intentional and its risk is accepted.