Bug Description
Arcane can restart itself, but cannot redeploy itself when it is managed as its own compose project.
On redeploy, Arcane detects the target as a compose-managed container and switches to project-based redeploy, then immediately shuts itself down and never comes back automatically. Manual docker compose up -d is required to recover.
Steps To Reproduce
Environment
- Arcane version:
v1.17.4
- Install method: Docker Compose
- Database: SQLite
- Projects directory:
/opt/docker/stacks
- Arcane is included under the projects directory and is managed as its own compose project
- Reverse proxy is used in front of Arcane, but this issue appears to be specifically about self-redeploy rather than proxy access
Compose setup
services:
arcane:
image: ghcr.io/getarcaneapp/arcane:1.17.4
container_name: arcane
restart: always
ports:
- "${ARCANE_PORT}:3552"
volumes:
- /var/run/docker.sock:/var/run/docker.sock
- ${ARCANE_DATA_DIR}:/app/data
- /opt/docker/stacks:/opt/docker/stacks
environment:
APP_URL: ${APP_URL}
ENCRYPTION_KEY: ${ENCRYPTION_KEY}
JWT_SECRET: ${JWT_SECRET}
PUID: ${PUID}
PGID: ${PGID}
LOG_LEVEL: ${LOG_LEVEL}
LOG_JSON: ${LOG_JSON}
OIDC_ENABLED: ${OIDC_ENABLED}
DATABASE_URL: ${DATABASE_URL}
UI_CONFIGURATION_DISABLED: "true"
PROJECTS_DIRECTORY: /opt/docker/stacks
DISK_USAGE_PATH: /opt/docker/stacks
networks:
- apps
healthcheck:
test: ["CMD-SHELL", "curl -fsS http://127.0.0.1:3552/api/health || exit 1"]
interval: 10s
timeout: 5s
retries: 3
start_period: 10s
networks:
apps:
external: true
name: apps
Steps to reproduce
- Install Arcane with Docker Compose
- Set the Projects Directory to
/opt/docker/stacks
- Move the Arcane compose project itself into that directory (for example
/opt/docker/stacks/arcane)
- Let Arcane discover and manage its own compose project
- Open the Arcane project page for the
arcane service
- Click Redeploy
Observed logs
Sanitized logs:
arcane | {"level":"INFO","msg":"RedeployContainer: detected compose container, using project-based redeploy","project":"arcane","service":"arcane"}
arcane | {"level":"INFO","msg":"Received shutdown signal"}
arcane | {"level":"INFO","msg":"Received interrupt signal. Shutting down…"}
arcane | {"level":"INFO","msg":"Scheduler stopped"}
arcane | {"level":"ERROR","msg":"Server forced to shutdown","error":"context deadline exceeded"}
arcane | {"level":"ERROR","msg":"Failed to run Arcane","error":"failed to run services: context deadline exceeded"}
After that, Arcane is no longer available until I manually run:
Then Arcane starts normally again.
Additional related logs after manual recovery:
arcane | {"level":"INFO","msg":"Arcane is starting","version":"v1.17.4"}
arcane | {"level":"INFO","msg":"Resolved database migration state","provider":"sqlite","currentVersion":45,"requiredVersion":45,"dirty":false}
arcane | {"level":"INFO","msg":"Database schema is up to date","provider":"sqlite","migrationVersion":45,"dirty":false}
arcane | {"level":"INFO","msg":"Database initialized successfully"}
arcane | {"level":"INFO","msg":"Starting HTTP server","addr":":3552","port":"3552","tls_enabled":false}
Notes
- This does not appear to be a general startup issue, because Arcane starts fine when manually brought up with
docker compose up -d
- This does not appear to be a simple restart issue, because Restart works
- The issue seems specific to self-redeploy when Arcane manages its own compose project
- I also see repeated warnings like this, though I am not sure whether they are directly related:
WARNING: Error loading config file: open /root/.docker/config.json: permission denied
Question
Is self-redeploy of Arcane currently unsupported, or is this a bug in the project-based redeploy flow when the target project is Arcane itself?
Expected Behavior
When Arcane manages itself as a compose project, clicking Redeploy should recreate and start the new Arcane container successfully, or at minimum complete the redeploy flow without requiring manual recovery.
Actual Behavior
- Restart works
- Redeploy stops Arcane
- Arcane does not come back automatically
- Manual
docker compose up -d is required to recover
Screenshots
No response
Arcane Version
v1.17.4
Installation Method
Docker Compose (Recommended)
Environment Type
Local Docker (Single instance)
Database Type
SQLite (Default)
Operating System
Linux
Docker Version
29.3.0
Browser & Version
No response
Relevant Logs or Error Messages
Docker Compose Configuration
Environment Configuration (.env file)
Additional Context
No response
Bug Description
Arcane can restart itself, but cannot redeploy itself when it is managed as its own compose project.
On redeploy, Arcane detects the target as a compose-managed container and switches to project-based redeploy, then immediately shuts itself down and never comes back automatically. Manual
docker compose up -dis required to recover.Steps To Reproduce
Environment
v1.17.4/opt/docker/stacksCompose setup
Steps to reproduce
/opt/docker/stacks/opt/docker/stacks/arcane)arcaneserviceObserved logs
Sanitized logs:
After that, Arcane is no longer available until I manually run:
Then Arcane starts normally again.
Additional related logs after manual recovery:
Notes
docker compose up -dQuestion
Is self-redeploy of Arcane currently unsupported, or is this a bug in the project-based redeploy flow when the target project is Arcane itself?
Expected Behavior
When Arcane manages itself as a compose project, clicking Redeploy should recreate and start the new Arcane container successfully, or at minimum complete the redeploy flow without requiring manual recovery.
Actual Behavior
docker compose up -dis required to recoverScreenshots
No response
Arcane Version
v1.17.4
Installation Method
Docker Compose (Recommended)
Environment Type
Local Docker (Single instance)
Database Type
SQLite (Default)
Operating System
Linux
Docker Version
29.3.0
Browser & Version
No response
Relevant Logs or Error Messages
Docker Compose Configuration
Environment Configuration (.env file)
Additional Context
No response