Checks whether the consumption of the CPU is limited for a Kubernetes deployment.
When your deployment uses too much CPU other deployments on the same node could become unstable.
Ensures that your configuration prevents a Kubernetes deployment to consume more memory than configured.
When your deployment uses too much memory other deployments on the same node could become unstable.
Verifies that at least two PODs of the same deployment exists.
It is hard to make single source of failure (like a single POD) resilient.
Ensures that a probe is configured so that Kubernetes is able to detect whether the POD is still working or it needs to be restarted.
An unresponsive POD may lead to no proper load balancing of requests. Sometimes, simple restart can help.
Tests the configuration of a rediness probe so that Kubernetes knows when a new starting POD is ready to receive traffic.
A POD that is still starting up may not be ready to accept traffic yet and thus lead to no proper load balancing of requests.
Verifies whether Kubernetes is configured to prevent scheduling traffic immediately after the
readinessProbe is successful.
May help to ensure that the newly started POD is stable enough to process traffic
Checks that the
imagePullPolicy is enforced by setting it to
always to ensure that always the latest version is running.
If set otherwise the OCI runtime may execute an older cached version of the image. See also weakspot "image latest tag".
Prevents usage of the
latest tag as a version.
latest version makes it hard to identify which version is running and may also lead to different versions running in the same cluster.
Enforces the deployment strategy
Prevents downtime while updating a deployment.
Horizontal POD Autoscaler
Verifies that a horizontal pod autoscaler is configured to schedule dynamically additional PODs.
Ensure that your system is properly scaled aligned to resource consumption.
Checks that your deployment is distributed across your cluster's nodes.
Just having multiple PODs (see "single replica") isn't enough in case all the PODs are running on a single node.
Tests that your Kubernetes cluster is running on multiple nodes.
A Kubernetes cluster with a single node is hardly resilient against node outages.