A newly deployed microservice on GKE is crashing sporadically due to resource constraints. The operations team needs to be immediately notified when…

This solution leverages the native integration between Google Kubernetes Engine (GKE) and the Google Cloud Operations suite. GKE automatically sends container and system logs to Cloud Logging and performance metrics to Cloud Monitoring. An Out-of-Memory (OOM) event is a system-level event captured in these logs. By creating a log-based metric in Cloud Monitoring, you can count occurrences of OOM log entries. Subsequently, an alerting policy can be configured to trigger a notification when this metric (or a standard metric like CPU utilization) crosses a defined threshold. This provides a fully automated, real-time notification system for both specified conditions, directly addressing the operational need.

A. This approach is manual and inefficient. kubectl is for direct interaction, not automated alerting, and ConfigMaps are for storing configuration data, not defining monitoring thresholds.

B. Security Command Center is a security management and threat detection service. It is not designed for monitoring application performance metrics like CPU utilization or OOM errors.

C. GKE clusters with Cloud Operations for GKE enabled (the default) already have a managed agent that collects logs and metrics. Deploying a custom agent is redundant and adds unnecessary management overhead.

1. Google Cloud Documentation - Overview of Cloud Operations for GKE: "When you create a GKE cluster, Cloud Logging and Cloud Monitoring are enabled by default... This includes support for GKE-specific logs and metrics, log-based metrics, and event-based alerting." This confirms the native integration used in option D.

Source: Google Cloud Documentation, "Overview of Cloud Operations for GKE", Section: "How Cloud Operations for GKE works".

2. Google Cloud Documentation - Log-based metrics: "Log-based metrics are Cloud Monitoring metrics that are based on the content of log entries... For example, you can create a log-based metric to count the number of log entries that contain a particular error message [like an OOM event]." This supports the method for tracking OOM errors.

Source: Google Cloud Documentation, "Log-based metrics overview", Introduction paragraph.

3. Google Cloud Documentation - Creating metric-based alert policies: "To be notified when a metric's value meets a condition that you specify, create a metric-based alerting policy." This explains how to create alerts for both standard metrics (CPU utilization) and the log-based metric created for OOM events.

Source: Google Cloud Documentation, "Alerting", Section: "Creating metric-based alert policies".

4. Google Cloud Documentation - Troubleshooting GKE: "If a container is terminated because it has reached its memory limit, this is considered an OOM kill event... You can find evidence of OOM kill events by running kubectl describe pod... or by looking for OOM kill events in your cluster's logs." This confirms that OOM events are logged, making them suitable for capture by Cloud Logging.

Source: Google Cloud Documentation, "Troubleshooting common issues in GKE", Section: "Troubleshooting resource issues".