Host and Docker metrics

The Squasher Collector is a small service that sends host metrics, and optionally Docker container metrics, from a Linux host to Squasher. The Squasher CLI plans, renders, installs, and verifies it.

Prerequisites

A Debian, Ubuntu, or RPM-based Linux host on amd64 or arm64, with systemd and sudo.
The Squasher CLI, signed in with squasher login.
An ingest key and the project ID. See Create an ingest key.
For container metrics: Docker, with its socket at /var/run/docker.sock.

Install the Collector

Preview the setup

The preview writes nothing and changes nothing on the host.

squasher infra plan --project <project-id> --service-name payments-host --containers

Leave out --containers if you do not need Docker metrics. --environment defaults to production, and --interval-seconds defaults to 60.

Render the install bundle

squasher infra render --project <project-id> --service-name payments-host --containers \
  --output ./squasher-infra

The new directory holds manifest.json, config.yaml, install.sh, and README.md. It contains no key. The command does not overwrite a directory that exists.

Install it

Review the files, then pass the key only to the install process:

export SQUASHER_API_KEY=<your ingest key>
sudo --preserve-env=SQUASHER_API_KEY ./squasher-infra/install.sh
unset SQUASHER_API_KEY

The installer writes the key to a file with mode 0600 and runs the Collector as the squasher-collector user.

Check it

sudo squasher infra doctor
squasher infra verify --project <project-id> --service-name payments-host --containers

verify returns healthy when CPU, memory, filesystem, optional Docker, and Collector health metrics reached Squasher in the last 180 seconds. Otherwise it returns incomplete and names the missing signal.

What the Collector sends

Group	Metrics
Host	CPU time and utilization, load average, memory, disk I/O, filesystem, network, paging, process count, uptime
Containers	CPU, memory usage and limit, block I/O, and network, for each Docker container
Collector	Its own CPU, memory, uptime, queue, and sent and failed metric points

The Collector does not read process command lines, environment variables, application logs, or eBPF data. For application logs and traces, use an SDK or OpenTelemetry. To see the data, select Infrastructure in the sidebar in Squasher.

Manage the Collector

Command	What it does
`squasher infra status`	Shows the local service state, memory, and CPU time
`sudo squasher infra doctor`	Checks the service, configuration, key file permissions, and Docker access
`squasher infra collectors list --project <project-id>`	Lists the Collectors seen in the last 24 hours
`squasher infra collectors overview --project <project-id>`	Counts active, degraded, and offline Collectors
`sudo squasher infra uninstall --yes`	Removes the Squasher-managed service, configuration, and key

Limits

The service runs with MemoryHigh=32M, MemoryMax=64M, and CPUQuota=10%.
Docker mode adds the Collector user to the Docker socket group.
The CLI renders a Linux systemd install only. For Kubernetes, see Kubernetes.

Next steps

Docker

Collect container logs with an OpenTelemetry Collector.

Kubernetes

Collect pod logs and workload metrics with Helm.

Get started

Guides

Host and Docker metrics

Prerequisites

Install the Collector

What the Collector sends

Manage the Collector

Limits

Next steps

Docker

Kubernetes

​Prerequisites

​Install the Collector

​What the Collector sends

​Manage the Collector

​Limits

​Next steps

Docker

Kubernetes

Prerequisites

Install the Collector

What the Collector sends

Manage the Collector

Limits

Next steps