Since Containly 2.7.3 Containly exposes /metrics for Prometheus: endpoint up, host CPU, memory and disk, container state, health, restarts, CPU and memory. Free, off by default, protected by a metrics token. Scrape config, metric list and Grafana queries.

Containly 2.7.3 can export its host and container metrics in the Prometheus text format. One scrape covers every host Containly manages. It is Free and off by default.

Switch it on

As an admin open Settings > Integrations.
Switch on Prometheus export. Leave Require an API token on (recommended).
Open Profile > API tokens, create a token with the permission Prometheus metrics (and nothing else) and copy it.
Copy the scrape config from the card into your prometheus.yml and put the token into the file named in credentials_file.

scrape_configs:
  - job_name: containly
    scrape_interval: 60s
    scheme: https
    metrics_path: /metrics
    authorization:
      type: Bearer
      credentials_file: /etc/prometheus/containly.token
    static_configs:
      - targets: ['containly.example.com']

Containly samples once per minute, so scrape_interval: 60s is enough; shorter intervals simply return the same values (a scrape never asks Docker, it reads Containly's cache).

Check it by hand:

curl -H "Authorization: Bearer $TOKEN" https://containly.example.com/metrics

Without the export switched on /metrics answers 404, without a token 401, with a token lacking the metrics permission 403. Only if /metrics can be reached from a trusted network alone, you may switch off Require an API token: then anyone who reaches Containly can read container names and resource usage.

Metrics

Labels: endpoint (host name in Containly), stack (Compose project, empty for standalone containers), container (name), image (reference without digest). No container IDs, so recreating a container does not create new series; at most 500 containers per host are listed.

Metric	Meaning
`containly_info{version,edition}`	Version and edition (free or pro)
`containly_endpoint_up{endpoint}`	1 when the host is reachable
`containly_host_cpu_percent`	Host CPU usage in percent
`containly_host_memory_used_bytes`, `containly_host_memory_total_bytes`	Host memory
`containly_host_disk_used_bytes`, `containly_host_disk_total_bytes`	Docker data disk
`containly_host_sample_timestamp_seconds`	Time of the last host sample
`containly_containers{endpoint,state}`	Number of containers per state
`containly_container_state{...,state}`	1 for the current state (running, exited, restarting, paused, created, dead)
`containly_container_health{...,health}`	1 for healthy, unhealthy, starting or none (no health check)
`containly_container_restarts`	Docker restart count (starts again at 0 when the container is recreated)
`containly_container_cpu_percent`	CPU of a running container in percent of one CPU
`containly_container_memory_bytes`	Memory of a running container
`containly_containers_dropped{endpoint}`	Containers left out above 500 per host

Grafana

Add Prometheus as a data source, then build panels with these queries:

Hosts down: containly_endpoint_up == 0
Host CPU: containly_host_cpu_percent
Host memory in percent: 100 * containly_host_memory_used_bytes / containly_host_memory_total_bytes
Disk in percent: 100 * containly_host_disk_used_bytes / containly_host_disk_total_bytes
Unhealthy containers: containly_container_health{health="unhealthy"} == 1
Not running: containly_container_state{state!="running"} == 1
Top CPU: topk(10, containly_container_cpu_percent)
Memory per stack: sum by (endpoint, stack) (containly_container_memory_bytes)
Restarts in the last hour: delta(containly_container_restarts[1h]) > 0

An alert rule example:

groups:
  - name: containly
    rules:
      - alert: ContainerUnhealthy
        expr: containly_container_health{health="unhealthy"} == 1
        for: 5m
        labels: {severity: warning}
        annotations:
          summary: "{{ $labels.container }} on {{ $labels.endpoint }} is unhealthy"

Privacy

The export is inbound only: Prometheus connects to Containly, Containly sends nothing anywhere. In the public demo the export is shown switched off.

Prometheus and Grafana

Switch it on

Metrics

Grafana

Privacy