Watch Docker Compose state and health in CTK

01

1. Prepare the Compose project

Start Docker and use your normal workflow to start the project’s Compose services. Connect that project folder in CTK. A Developer or Team plan must include Docker and have room for the observed services.

CTK associates containers through Compose project-directory information. A standalone container or a Compose project launched against another directory is not automatically part of the folder you selected.

02

2. Start the Docker Compose watch

Choose Start watching, select Docker Compose and read the method description. Submit with Start watching. This action observes the containers; it is not a deployment or startup command.

In Observe, check the service and container identity in How before reading the state. Similar container names across different projects are not enough to confirm you selected the right workload.

03

3. Read state and health separately

Running describes the container process. Healthy describes a successful configured health check. Without a health reading, CTK cannot establish readiness just because the process has not exited.

A restarting container, an exit code, a Docker error or an out-of-memory report provides different evidence. Read the actual Why field rather than interpreting every restart as a memory failure. A failing health check identifies a failed check, not automatically its underlying defect.

04

4. Follow a change through history

Use Timeline for the repository’s recorded changes and Activity for the workspace view. Compare the failure and the later reading by service and time. Repeated unchanged polling is not a new incident on every refresh.

Alarms reflects currently stated causes in watched repositories. A previous failure can leave Alarms after recovery while its recorded change remains in history. That is expected, not evidence that the earlier failure was erased.

05

5. Distinguish a stale snapshot from a stopped app

If Docker becomes unavailable, CTK can retain the last successful snapshot and show a connection interruption. Read the timestamp; old evidence is not a new check that the application is healthy or down.

Restore Docker access outside CTK, then wait for a successful observation. Stop watching when you no longer want collection. This releases the watch’s service usage without stopping the Compose workload.

06

What success looks like

The intended Compose services appear with their observed state and any health checks they actually expose. You can distinguish current evidence, missing health evidence and an interrupted read without guessing that they mean the same thing.