Containers: It's what's inside that counts. And container monitoring, the technology of peering into running Docker containers to find out what's going on within, helps you get a leg up. As Docker's own APIs for monitoring and introspection have matured, so has the latest wave of container-monitoring tech that leverages them.
Here are six apps and services that span the gamut of what's going on in monitoring, from simple open source tooling to sophisticated monitoring that's part of a holistic, enterprise-spanning solution.
Google's answer to the container introspection problem is cAdvisor, a utility that itself is crated in a Docker container and generates quick, useful information on the basic behaviors of running containers. It's simple to deploy and use, perhaps to a fault -- you can monitor only other containers running on the same host, so it's not as useful for multinode deployments. Plus, any long-term aggregation or distillation of statistics is entirely up to you.
Brewed up in the labs of Soundcloud, Prometheus.io is an open source framework for systems monitoring and alerting. Its power lies in how it can ingest data from a colossal spectrum of sources, including containers. Even better, the data sources for containers aren't limited and cover the likes of cAdvisor, Kubernetes, and CoreOS's Etcd.
A sideways step from cAdvisor and Prometheus is Scout, a monitoring service rather than a stand-alone open source project. It covers far more territory than cAdvisor alone -- namely, it aggregates data from across multiple hosts and has a plug-in architecture allowing for a broad range of integrations. But it's also a for-pay service at a cost of $10 per host, so those seeking an entirely free solution are bound to be let down.
Also in the same vein of "container monitoring as a service" is DataDog, which comes in at $15 per host, but with more detailed and flexible reporting features than Scout. (See the rundown from Rancher Labs comparing Scout and DataDog.)
Many of the folks getting into the Docker-monitoring game are longtime system monitoring mavens. Logentries is among them, and the company seems to have sensed which way the wind's blowing with containers. It recently rolled out a free logging service for Docker containers, which hooks into a few common Docker logging APIs (log messages, container metrics, container detection) exposed by way of Docker 1.5 and up. A 30-day trial of Logentries' other (for-pay) services is included, but the container logging service itself remains free. Minor caveat: If you run Docker in a restricted environment, such as Google Compute Engine, you'll need to run the Logentries container in privileged mode.
An in-cloud and on-prem monitoring solution with several integrations, Sematext recently added Docker monitoring to its SPM solution. Like others in this vein, the monitoring is conducted via a container, with events piped back out to a variety of sources, and one agent can be used to monitor multiple servers. Also included is monitoring for container-level events (stops, starts, and so on) and managing the (sometimes copious) logs generated by containers. Plans start at 3.5 cents per server-hour.