GCP Managed Instance Group Not Respecting Auto-Scaling Policies During High Load
I'm sure I'm missing something obvious here, but I'm working with an scenario where my GCP Managed Instance Group (MIG) is not scaling up as expected during peak load periods, despite having configured auto-scaling policies based on CPU utilization. My current settings have a target CPU utilization of 60% and a minimum of 2 instances and a maximum of 10 instances. However, when the CPU usage spikes to over 80%, it seems like the MIG doesn't scale up until it hits a much higher threshold, and sometimes even then, it takes an unusually long time to add instances. I've tried adjusting the cooldown periods and verified that the health checks are passing. Here’s the relevant configuration for my auto-scaling policy: ```json { "autoscalingPolicy": { "coolDownPeriodSec": 60, "minNumReplicas": 2, "maxNumReplicas": 10, "cpuUtilization": { "utilizationTarget": 0.6, "aggPolicy": "GAUGE" } } } ``` The instances are running on n1-standard-1 machine types, and I have enabled Stackdriver Monitoring to get more insights. When I check the monitoring dashboard, the CPU usage is indeed spiking, but the MIG is still holding at 2 instances. I’ve also looked at the logs, but there are no errors showing that the scaling policies are failing. Could it be a configuration scenario or is there a known limitation with auto-scaling under heavy loads? Any suggestions on how to debug this further would be greatly appreciated! This is part of a larger application I'm building. What's the best practice here? My development environment is Windows. Any ideas what could be causing this?