Troubleshooting

Each entry follows symptom → check → fix, and every command says which server it runs on. To start the installation over from scratch, see Upgrade and uninstall.

The dashboard does not open

Symptom: http://<dashboard server IP>:3001 does not load in the browser, or shows an error.

Check (on the dashboard server, in the everyup directory):

docker compose ps                # STATUS should be healthy
docker logs everyup --tail 100   # look for startup errors
curl -fsS http://localhost:3001/api/v1/health

Fix:

• If curl succeeds but the browser does not, open port 3001 in the firewall or security group.
• If the container keeps restarting, read the first error in the logs. A wrong environment variable value is the usual cause; see Web configuration.

The installer stops on the Compose version

Symptom: the install command ends with Docker Compose v2.23.1 or newer is required.

Check (on the monitored server):

docker compose version

Fix: update the Docker Compose plugin to 2.23.1 or newer. If Docker came from your package manager, updating the docker-compose-plugin package usually does it. The standalone docker-compose (v1) is not supported.

The installer cannot reach Docker

Symptom: the install command ends with Docker Engine is not reachable. or /var/run/docker.sock is not readable.

Check (on the monitored server):

sudo docker info --format '{{.ServerVersion}}'
ls -l /var/run/docker.sock

Fix:

• Run the install command with sudo sh. Make sure you copied the command exactly as Web showed it.
• If sudo docker info fails, check the error for socket permissions and daemon connectivity. If the daemon is stopped, start it with sudo systemctl start docker.

Installation fails with a connection code error

Symptom: Exchanging the one-time EveryUp join code... is followed by curl: (22) The requested URL returned error: 401.

Fix: a connection code works once, within 10 minutes of being issued. Click 새 코드 (new code) on the install screen to generate a new command and run it. Re-running the old command gives the same error.

The monitored server cannot reach Web

Symptom: the install command ends with curl: (7) Failed to connect or curl: (28) … timed out.

Check (on the monitored server, with the Web address from the install command):

curl -fsS https://<Web address>/api/v1/health

Fix:

• If it cannot connect, open port 3001 (or your reverse proxy port) in the dashboard server's firewall or security group.
• If the address in the command is localhost or internal-only, restart Web with EVERYUP_PUBLIC_URL and generate a new install command.

The Docker environment never reaches 수집 중 (collecting)

Symptom: installation finished, but the environment stays at 설치 대기 (waiting for install) on the Docker 환경 screen.

Check (on the monitored server):

sudo docker compose --env-file /opt/everyup-agent/.env -f /opt/everyup-agent/compose.yaml ps
sudo docker logs everyup-agent --tail 50
sudo grep EVERYUP_WEB_BASE_URL /opt/everyup-agent/compose.yaml

Fix: the Collector connects to Web through EVERYUP_WEB_BASE_URL, so the address must be reachable from inside the Collector container. Even on the same server, localhost inside the container points to the Collector itself, not Web. Change it to a Compose service name or a host-reachable IP, then restart:

sudo docker compose --env-file /opt/everyup-agent/.env -f /opt/everyup-agent/compose.yaml up -d

The Docker Collector cannot read the Docker socket

Symptom: the everyup-agent logs show query docker socket … permission denied and no services appear.

Fix: the one-line installer detects the Docker socket group ID and writes EVERYUP_DOCKER_GID automatically. For a manual deployment, set that value to stat -c '%g' /var/run/docker.sock and add it through group_add. Use user: "0:0" only as a short-lived diagnostic fallback. To narrow socket access in production, use the Docker socket proxy guide.

Logs are not showing up

Symptom: the service appears, but its 로그 (Logs) tab is empty.

Check (on the monitored server):

docker logs <app container> --tail 20

Fix: send a request to the app and check docker logs again. If it is still empty, check the app and proxy log-output settings. If they only write to files inside the container, configure stdout/stderr output. This collection path cannot read content absent from Docker logs.