Frequently Asked Questions (FAQs)

This section contains the answers to the frequently asked questions (FAQ).

Why do I occasionally see fewer pods in the system metrics for public helm charts?

In public helm charts, the limits and requests for resources are determined by the user configuring them. You can manage these settings in the Helm chart YAML configuration. When set, Kapstan can accurately assess your system metrics for the pods. If not set, it cannot calculate them, resulting in missing data in the system metrics section.

Why is using the latest image tag not recommend?

Using the latest image tag for applications is not recommended because it is a moving tag, meaning that the image "latest" refers to can change. This makes it difficult to determine what application logic is currently deployed when inspecting your live environment. By using a fixed tag like "1.1.0" instead, it is always overtly clear what version of your application is currently running.

Additionally, moving tags can cause subtle issues with Kubernetes' image pull policy, which can lead to unexpected behavior in your deployments. When using "latest", Kapstan will set your pod's pull policy to Always in order to ensure that when the tag is moved, the new image is always pulled. However, this can cause issues when the tag is moved if the application is not immediately redeployed.

For example, say you have a deployment running with the latest tag and this refers to application version 0.0.1. You update your code and build a new image for version 0.0.2 and move the latest tag to point to this. At this point, any new pods created (ex: from autoscaling) will pull the new image with application version 0.0.2. However, any existing pods will not be updated and will still be running 0.0.1.

This can lead to a situation where you have multiple versions of your application running in your cluster at the same time, and it is unclear from an outside perspective.