Container

Supported OS Linux Windows

Integration version1.0.0

Overview

This check reports a set of metrics about any running containers, regardless of the runtime used to start them.

NOTE: The container check is different from the containerd check. The container checks report standardized metrics for all containers found on the system, regardless of the container runtime. The containerd is dedicated to containerd runtime and publishes metrics in the containerd.* namespace.

Setup

Installation

Container is a core Datadog Agent check and is automatically activated if any supported container runtime is detected. Configuring access to supported container runtimes (Docker, containerd) may be required depending on your environment.

Installation on containers

The container check requires some folders to be mounted to allow for automatic activation. This is handled by the official Helm Chart, the Datadog Operator, and as documented set ups for Kubernetes, Docker, ECS, and ECS Fargate.

Configuration

The container check exposes specific configuration settings to emit extended memory metrics in addition to standard set. To customize common fields or to force the activation of the container check, follow these steps:

Create the container.d/conf.yaml file in the conf.d/ folder at the root of your Agent’s configuration directory.
Restart the Agent

The container check can collect metrics about CPU, Memory, Network and Disks IO. Some metrics may not be available depending on your environment (Linux / Windows, for instance).

Validation

Run the Agent’s status subcommand and look for container under the Checks section.

Data Collected

Metrics


container.cpu.limit (gauge)	The maximum CPU time available to the container Shown as nanocore
container.cpu.system (gauge)	The container system CPU usage Shown as nanocore
container.cpu.throttled (gauge)	The total cpu throttled time Shown as nanosecond
container.cpu.throttled.periods (gauge)	The number of periods during which the container was throttled
container.cpu.usage (gauge)	The container total CPU Usage Shown as nanocore
container.cpu.user (gauge)	The container userspace CPU usage Shown as nanocore
container.io.read (gauge)	The number of bytes read from disks by this container Shown as byte
container.io.read.operations (gauge)	The number of read operations done by this container
container.io.write (gauge)	The number of bytes written to disks by this container Shown as byte
container.io.write.operations (gauge)	The number of write operations done by this container
container.memory.active_anon (gauge)	The container active anonymous memory (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.active_file (gauge)	The container active file-backed memory (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.cache (gauge)	The container cache usage Shown as byte
container.memory.commit (gauge)	The container commit memory usage Shown as byte
container.memory.commit.peak (gauge)	The container peak commit memory usage Shown as byte
container.memory.file_dirty (gauge)	The container memory that are waiting to get written back to the disk (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.file_mapped (gauge)	The container mapped file memory (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.file_writeback (gauge)	The container memory that is queued for syncing to disk (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.inactive_anon (gauge)	The container inactive anonymous memory (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.inactive_file (gauge)	The container inactive file-backed memory (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.kernel (gauge)	The container kernel memory usage Shown as byte
container.memory.limit (gauge)	The container memory limit Shown as byte
container.memory.major_page_faults (count)	Number of major page faults incurred
container.memory.oom_events (gauge)	The number of OOM events triggered by the container
container.memory.page_faults (count)	Total number of page faults incurred
container.memory.page_tables (gauge)	The container memory allocated for page tables (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.refault_anon (count)	The number of refaults of previously evicted anonymous pages (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.refault_file (count)	The number of refaults of previously evicted file pages (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.rss (gauge)	The container RSS usage Shown as byte
container.memory.shmem (gauge)	The container swap-backed memory such as tmpfs (requires extended_memory_metrics flag in the check config), shm segments… Shown as byte
container.memory.soft_limit (gauge)	The container memory soft limit Shown as byte
container.memory.swap (gauge)	The container swap usage Shown as byte
container.memory.unevictable (gauge)	The container memory that can’t be reclaimed (requires extended_memory_metrics flag in the check config) Shown as byte
container.memory.usage (gauge)	The container total memory usage Shown as byte
container.memory.usage.peak (gauge)	The maximum memory usage recorded since the container started Shown as byte
container.memory.working_set (gauge)	The container working set usage Shown as byte
container.net.rcvd (gauge)	The number of network bytes received (per interface) Shown as byte
container.net.rcvd.packets (gauge)	The number of network packets received (per interface)
container.net.sent (gauge)	The number of network bytes sent (per interface) Shown as byte
container.net.sent.packets (gauge)	The number of network packets sent (per interface)
container.pid.open_files (gauge)	The number of open file descriptors (Linux only)
container.pid.thread_count (gauge)	The number of threads running inside this container
container.pid.thread_limit (gauge)	The maximum number of threads for this container
container.restarts (gauge)	The number of container restarted
container.uptime (gauge)	The container uptime Shown as second

If you need extended container memory metrics extended_memory_metrics has to be set to true. It will enable following metrics: active_anon, inactive_anon, active_file, inactive_file, unevictable, shmem, file_mapped, file_dirty, file_writeback, page_tables, refault_anon, refault_file

Troubleshooting

Need help? Contact Datadog support.