CRI is a core Datadog Agent check that needs to be configured in the datadog.yaml with the cri.d/conf.yaml.
In datadog.yaml, configure your cri_socket_path for the Agent to query your current CRI (you can also configure default timeouts). In cri.d/conf.yaml, configure the check instance settings such as collect_disk if your CRI (such as containerd) reports disk usage metrics.
Note: If you’re using the Agent in a container, set the DD_CRI_SOCKET_PATH environment variable to automatically enable the CRI check with the default configuration.
Installation on containers
If you are using the Agent in a container, setting the DD_CRI_SOCKET_PATH environment variable to the CRI socket automatically enables the CRI integration with the default configuration.
For example, to install the integration on Kubernetes, edit your daemonset to mount the CRI socket from the host node to the Agent container and set the DD_CRI_SOCKET_PATH env var to the daemonset mountPath:
Note: The /var/run directory must be mounted from the host to run the integration without issues.
Configuration
Edit the cri.d/conf.yaml file, in the conf.d/ folder at the root of your Agent’s configuration directory to start collecting your crio performance data. See the sample cri.d/conf.yaml for all available configuration options.
Run the Agent’s status subcommand and look for cri under the Checks section.
Data Collected
Metrics
CRI collect metrics about the resource usage of your containers running through the CRI.
CPU and memory metrics are collected out of the box and you can additionally collect some disk metrics if they are supported by your CRI (CRI-O doesn’t support them).
cri.cpu.usage (gauge)
Cumulative CPU usage (sum across all cores) since object creation Shown as nanocore
cri.disk.inodes (gauge)
Represents the inodes used by the images Shown as inode
cri.disk.used (gauge)
Represents the bytes used for images on the filesystem Shown as byte
cri.mem.rss (gauge)
The amount of working set memory in bytes Shown as byte
cri.uptime (gauge)
Time since the container was started Shown as second