Skip to main content

Collect Diagnostics

Quick checks​

Capture: time, operation, affected app/domain, expected result, actual result, and error.

On the Swarm manager:

docker service ls
docker service ps captain-captain --no-trunc
docker service logs captain-captain --tail 100

Check by symptom:

  • App: Find its name in docker service ls. Older apps may use srv-captain--APP_NAME. Use docker service ps SERVICE_NAME and docker service logs SERVICE_NAME.
  • Networking: Check DNS and test HTTP and HTTPS separately.
  • Storage: Check mounts, disk space, and node placement.

Collect a full server report​

Use the diagnostic collector if the dashboard is unavailable, several services fail, or quick checks are inconclusive. Run on a Debian/Ubuntu Swarm manager with systemd:

curl -fL https://raw.githubusercontent.com/caprover/caprover-website/master/scripts/caprover-diagnostic.sh -o caprover-diagnostic.sh
less caprover-diagnostic.sh
sudo sh ./caprover-diagnostic.sh

Output: /tmp/caprover-diagnostic.*.txt, with owner-only permissions.

Checks include:

  • Port owners (80/443/3000), host web servers, local HTTP responses
  • Docker/Swarm services, task errors, specifications, and logs
  • /captain mounts and permissions, networks, firewall, kernel and daemon errors
  • Package history and available container runtimes

The collector briefly tests local hello-world and nginx:1.27.2 images if already available. It does not pull images or change CapRover service configuration.

Read the report summary first. Examples:

  • Port 80/443 owned by another process: CapRover NGINX may fail to bind.
  • Swarm task rejected: Inspect its task error.
  • /captain mount missing: Inspect the host path and mounts.
  • Docker API warning: Check other evidence too. In issue #2377, host NGINX occupied port 80.

Before sharing​

  • Review the entire report for passwords, tokens, private user data, app names, and identifying host details.
  • The collector omits Docker environment variables and CapRover configuration contents, but logs, arguments, and package history can still contain sensitive data.
  • Share only relevant, sanitized sections.

See Server Diagnostics for targeted recovery steps.

Restart or inspect services​

docker service update captain-captain --force   # Restart CapRover
docker service update SERVICE_NAME --force      # Restart an app
docker service logs SERVICE_NAME --since 10m

Run Docker commands on a Swarm manager. Get the exact physical service name from docker service ls. For a shell inside an app, use docker exec on the node hosting its task (see below).

Diagnostics​

Start with the failing request or app, its timestamp, and the exact error. In the dashboard, inspect the application's deployment history and logs. On the manager host:

docker service ls
docker service ps captain-captain --no-trunc
docker service logs captain-captain --tail 100
docker system df
  • For an app, find its service in docker service ls and inspect docker service ps SERVICE_NAME --no-trunc and docker service logs SERVICE_NAME --tail 100.
  • Newer apps generally use the app name; legacy apps may use srv-captain--APP_NAME.
  • For DNS or HTTPS failures, check records, port reachability, and the domain's certificate status.
  • For multi-node issues, verify node availability and private networking.
  • Remove passwords, tokens, and private data from logs before sharing them.
  • See Troubleshooting to find a guide by symptom.

For an outage involving several system services or an unclear host failure, collect a full server report on the manager.

Dashboard unavailable on port 3000​

  • On the manager, check docker service ps captain-captain --no-trunc, then inspect docker service logs captain-captain --since 60m and docker service logs captain-nginx --since 60m.
  • Run curl -v http://127.0.0.1:3000/ on the manager.
  • If the local request works but the public IP does not, inspect the provider and host firewalls and the port mapping.
  • If the CapRover service is restarting, use the task error and logs before changing networking.
  • To force a restart after correcting the underlying issue, run docker service update captain-captain --force.

Restart or inspect an app​

  • From the dashboard, save the app's configuration to request a service update.
  • For a shell in a running task, first obtain its physical service name from docker service ls, find the host using docker service ps SERVICE_NAME, and run these on that host:
docker ps --filter label=com.docker.swarm.service.name=SERVICE_NAME
docker exec -it CONTAINER_ID /bin/sh

Replace SERVICE_NAME and CONTAINER_ID with the values you inspected. Not every image contains /bin/bash; /bin/sh is more common. A container shell has broad access to the app's data and secrets.