As a Site Reliability Engineer, you support an application written in Go that runs on Google Kubernetes Engine (GKE) in production. After releasing a new version of the application, you notice the app

Increasing the CPU limit would address CPU starvation, but the problem description clearly indicates a memory issue (heap usage growth), not a CPU bottleneck.

Adding high memory compute nodes to the cluster provides more total memory capacity, but if the application's *deployment* memory limit is too low, it will still be OOMKilled regardless of the underlying node capacity.

The symptoms described (constant heap usage growth leading to restarts) clearly point to a memory issue, likely a memory leak or the application requiring more memory than allocated. Increasing the memory limit in the application's Kubernetes deployment provides the container with more RAM, preventing it from being OOMKilled (Out Of Memory Killed) by Kubernetes and thus stopping the restarts.

Cloud Trace is used for distributed tracing to understand latency and execution paths, which is valuable for performance analysis but not directly for mitigating a memory exhaustion issue causing restarts, especially when a profiler already identified heap growth.