Blog

Predictive failure alerts in the OpenShift cluster Alertmanager

Ruptura sends predicted and detected workload failures to the OpenShift cluster Alertmanager, so they follow the routes, silences and receivers your team already maintains.

The one replica that leaks while its siblings stay flat

Averages and global thresholds miss a single leaking replica and a release that leaks faster than the one before it. Here is how to catch both before the OOM kill.

Kubernetes OOMKilled (exit code 137): why it happens and how to see the next one coming

What exit code 137 and OOMKilled mean, the five real causes, how to find yours in five minutes, and how to get alerted before the next kill.

Predict Kubernetes OOM kills 30 minutes ahead with one PromQL alert

A copy-paste Prometheus alert that warns before a container is OOM killed, how predict_linear works, and where it breaks.