Monitor containers in the Orion Platform
This Orion Platform topic applies only to the following products:
SAM — VMAN
Similar to VMs, containers isolate applications and their dependencies into self-contained units that can run anywhere without interfering with each other. Unlike VMs that each have their own virtual OS and hardware, containers share the host system kernel with other containers to save space, increase efficiency, and improve developer productivity.
The Orion Platform supports monitoring for the following types of container environments:
- Docker Swarm 17.12 and later
- Kubernetes (K8s), including Microsoft Azure Kubernetes (AKS)
- Apache Mesos
The ability to monitor containers hosted on Linux nodes in the Orion Platform enables you to:
- Display details about your container infrastructure, including hosts, host clusters, environment dependencies, and deployments.
- Track container, host, and other infrastructure metrics to help ensure and plan capacity.
- View in-depth data about individual containers on the Container Details page, as shown here.
- Analyze container activity in the AppStack Environment.
- Organize containers on Orion Maps.
Container entities are also available in Performance Analysis (PerfStack) dashboards. For example, if you track TCP Port Monitor and Average Response Time metrics for an application on a server that hosts a containerized environment and you also have CPU utilization for different containers running throughout the history of the application's deployment, you can trace back to see what happened for containers running at different points in history.
To get started monitoring containers quickly:
- Adjust your environment to satisfy container monitoring requirements.
- Add Linux host servers as managed Orion nodes.
- Use the Add Container Service wizard to generate scripts for Linux host servers.
- Run provided scripts on Linux nodes, and then wait a few minutes for container status and performance metrics to appear in various areas of the Orion Web Console, such as the Container Details page. See Display container data in the Orion Web Console for details.
Note the following details about container monitoring:
- The default polling interval is five minutes, but you can adjust intervals globally. See Update an existing container service.
- Each managed Orion node that hosts containers consumes a license. To monitor all containers in a single cluster, add each server from the cluster as an Orion node.
- Starting in Orion Platform 2020.2, you cannot add a container service if FIPS mode is enabled. If containers were added before FIPS mode was enabled, remove them from nodes and then delete the container service. Otherwise, container polling will continue.
- In Orion Map subviews of Node Details pages, containers appear as authorized children of nodes.
- No out-of-the-box alerts and reports currently exist for container monitoring, but you can use related conditions and events to create your own. For example, set up an alert to trigger email when a container restarts more than five times. See Use alerts to monitor your environment.
When you add a container service, Orion credentials are stored in a plain text file that stays in memory for two hours. SolarWinds recommends creating a dedicated Orion user account to manage container services so critical credentials are not exposed.
Want to learn more about monitoring containers? Check out these videos:
- SolarWinds Lab Bits: Curious About Containers? (2:59)
- SolarWinds Lab Bits: Monitoring Containers in SAM and Virtualization Manager (11:55)
SolarWinds documentation describes how to display container data in the Orion Web Console. To learn about manipulating containers directly, refer to third-party documentation provided by the vendor. For example, to learn about swarm mode, see Docker docs (© 2020 Docker, Inc., available at docs.docker.com, obtained on October 26, 2020.)
SolarWinds AppOptics works well with ephemeral (short lived) containers and hosts, and includes integrations with Docker, Kubernetes, and Mesos. AppOptics APM agents install easily into containers, providing performance metrics for their containerized applications. Click here to learn more.