Container Monitoring

Container Monitoring

Monitoring containers differs from monitoring traditional servers in one key way: containers are ephemeral. A container might be recreated with a new ID at any time, making container-by-ID monitoring unreliable. Modern container monitoring uses labels and service names to track metrics across container restarts. The monitoring approach scales from simple (docker stats) to production-grade (Prometheus + cAdvisor + Grafana).

Docker built-in monitoring

# Live resource usage for all containers:
docker stats

docker stats output

CONTAINER ID   NAME    CPU %    MEM USAGE / LIMIT    MEM %   NET I/O         BLOCK I/O
abc123         web     0.52%    45.2MiB / 512MiB     8.83%   2.3GB / 456MB   12MB / 890MB
def456         db      1.24%    256.8MiB / 1GiB      25.1%   123MB / 67MB    890MB / 4.5GB

# Non-interactive (for scripting):
docker stats --no-stream --format "table {{.Name}}	{{.CPUPerc}}	{{.MemUsage}}"

# Inspect a specific container's resource usage:
docker inspect mycontainer | grep -A5 "Memory\|CpuShares"

Prometheus for containers

# cAdvisor exposes container metrics to Prometheus:
docker run -d   --name cadvisor   --privileged   -p 8080:8080   -v /:/rootfs:ro   -v /var/run:/var/run:ro   -v /sys:/sys:ro   -v /var/lib/docker/:/var/lib/docker:ro   gcr.io/cadvisor/cadvisor:latest

# Add cAdvisor to Prometheus scrape config:
# scrape_configs:
#   - job_name: 'cadvisor'
#     static_configs:
#       - targets: ['localhost:8080']

# Key metrics available:
# container_cpu_usage_seconds_total  — CPU usage per container
# container_memory_usage_bytes       — memory per container
# container_network_transmit_bytes_total — network per container

Container log management

# View container logs:
docker logs myapp                     # All logs
docker logs myapp --tail 100          # Last 100 lines
docker logs myapp -f                  # Follow (like tail -f)
docker logs myapp --since 1h          # Last hour
docker logs myapp --since "2025-06-09T14:00:00"

# Prevent log files from filling disk (set in /etc/docker/daemon.json):
# {
#   "log-driver": "json-file",
#   "log-opts": { "max-size": "100m", "max-file": "3" }
# }

# Ship container logs to syslog/Fluentd/ELK:
docker run --log-driver=syslog --log-opt syslog-address=tcp://logserver:514 myapp

Conclusion

For production container monitoring, the standard stack is cAdvisor (container metrics exporter) + Prometheus (metrics storage) + Grafana (dashboards). Import the Docker dashboard from Grafana's community library (ID 893 or 14282) to immediately get CPU, memory, network, and restart counts per container. Always configure log rotation in /etc/docker/daemon.json — a verbose application running for months without log rotation fills the disk and crashes the Docker host.