Monitoring overview

更新时间:
复制 MD 格式

ACK is compatible with Alibaba Cloud observability products such as CloudMonitor and Managed Service for Prometheus, and provides monitoring add-ons to help you observe cluster health and proactively identify and respond to issues. This topic describes the full-stack monitoring solutions for ACK clusters, covering basic resources, applications, clusters, events, control plane components, networks, and kernel-level containers.

Cluster observability

ACK provides the following monitoring modules.

Module

Description

References

Add-on

Basic resource monitoring

Monitor the usage and health of basic cluster resources, including CPU, memory, and network, with CloudMonitor or Prometheus. Enable alert notifications on key metrics to improve stability.

[Discontinued] Basic resource monitoring

metrics-server

Connect to and configure Managed Service for Prometheus

ack-arms-prometheus

Open source Prometheus monitoring

ack-prometheus-operator

Application monitoring

Use Application Real-Time Monitoring Service (ARMS) and ack-onepilot for topology analysis, API and event monitoring, tracing, and performance bottleneck detection in containerized applications.

Java application monitoring

ack-onepilot

Python application monitoring

Go application monitoring

Cluster monitoring

Use Application Monitoring eBPF Edition to collect container performance data without code intrusion, identify pod issues, pinpoint related Services and controller workloads, and reduce time to detection.

[Deprecated] Cluster topology monitoring

ack-arms-cmonitor

Event monitoring

Use Node Problem Detector (NPD) and the Kubernetes Event Center of Log Service (SLS) to monitor and notify on system status in real time, diagnose node exceptions and convert them to events, and manage alerts with closed-loop and offline notifications.

Event monitoring

ack-node-problem-detector

Control plane component monitoring

Monitor control plane components in real time with Prometheus and Grafana, including the API server, etcd, kube-scheduler, and kube-controller-manager. Optimize component access and configure self-managed Prometheus.

View control plane component dashboards

Kube API server

Kube Controller Manager

Cloud Controller Manager

Container Service for Kubernetes:kube-scheduler

etcd

Network monitoring

Integrate the Ingress log service and use Ingress Dashboard with ARMS for joint Ingress troubleshooting. Monitor and troubleshoot CoreDNS. Visualize network traffic and service topology in Terway-based clusters to achieve container network and service observability.

Ingress Dashboard

NGINX Ingress controller

Monitor the CoreDNS component

CoreDNS

Achieve network observability with ACK Terway and Cilium Hubble

Terway network plugin

Kernel-level container monitoring

Monitor containers at the OS kernel level with System Observer Monitoring (SysOM) to facilitate deployment and migration of containerized applications.

SysOM kernel-level container monitoring

ack-sysom-monitor