I have detected the ingress-nginx pods get OOMkilled from time to time, meaning they are trying to use more RAM than allowed in their resource limits.

As of this writing, the k8s deployment:

• mem request: 2G
• mem limit: 3G
• cpu request: 2
• cpu limit: 3

As of this writing, the dedicated worker virtual machine:

• has 8GB RAM, usually 4GB RAM free.
• has 4 CPU, usually idle

We can upscale the deployment to better use the resources of the virtual machine.