JobMonitoring

If monitoring has been enabled with cAdvisor, user can monitor live CPU/Mem/network usage of the container (job).

The web interface, while job is running, display a usage button that shows the current container usage.

When job is over, those metrics are not available anymore.

However, GoDocker provides Prometheus metrics. If available, one can get past metrics of a task on Prometheus web interface with query:

rate(container_cpu_usage_seconds_total{id="/docker/container_cpu_system_seconds_total{id="/docker/561b95090f96142c0a71912dba9b1f3d0768912f6ff4db1c2c6c1efc547a312d"}"} [5m])

• "/docker/XXX": is the name of the container. It can be found in the job details.
• container_cpu_usage_seconds_total: is a prometheus metric for CPU
• 5m: rate 5 minutes ago