How to effectively monitor abnormal occurrences of CPU and memory usage in containers using PrometheusQL expressions?

[yuan xiao]

Scenario 1: Monitoring sudden spikes in container memory usage, such as going from 300M in one minute to 11G in the next minute. How can we achieve effective monitoring and alerting in such cases?
Scenario 2: Monitoring continuous memory growth of containers, such as a sustained increase in container memory usage over the past 4 hours. How can we implement effective monitoring and alerting for this situation?

When either of the above two scenarios occur, Prometheus alerting rules can be triggered to send out alert notifications.

When either of the above two scenarios occur, Prometheus alerting rules can be triggered to send out alert notifications.The Prometheus scraping interval is 1 minute.We have deployed monitoring components such as Prometheus, Grafana, Alertmanager, Pushgateway、exporter and cAdvisor.