I did a PoC on GKE Autopilot and it turned to be very expensive for the few days it ran. From my troubleshooting I found that a few workloads did not have resource limits. My question is are there any guidelines to provision cost efficient GKE Autopilot clusters without compromising on Reliability? Interesting that GKE Autopilot is the default mode.
Also is there is checklist to evaluate whether my workload can be run on Autopilot or rather best fit to run on Autopilot clusters