Docker monitoring

This is a dashboard for monitoring Docker container metrics collected by cAdvisor and stored in Prometheus TSDB.

Inspired by 193 and 8321

Features:

• Depend on only cAdvisor metrics.
• Select Jobs, Nodes and Services. Multi-value and Select-all are supported.
• Use the new Time series panel. Faster
• Merge network transmit and receive into one longer panel.
• Merge Disk read and write into one longer panel.

This dashboard use the following metrics and their corresponding cadvisor model:

• container_cpu_user_seconds_total : cpu
• container_memory_usage_bytes : memory
• container_network_receive_bytes_total : network
• container_network_transmit_bytes_total : network
• container_fs_reads_bytes_total : diskIO
• container_fs_writes_bytes_total : diskIO

This dashboard expects the following labels on the metrics:

• job
• service
• node
• container

You may assign container_label_com_docker_swarm_service_name to service, and name to container using Prometheus’ metric_relabel_configs. Here is an example of the scrape rules.

Copy

metric_relabel_configs:
      - source_labels: ['container_label_com_docker_swarm_service_name']
        target_label: 'service'
      - source_labels: ['name']
        target_label: 'container'

You can add container_label_com_docker_swarm_service_name to the container by setting allowlisted_container_labels: ['com.docker.swarm.service.name'] for the cadvisor.