GhostCMS trigger for post published is not working when testing trigger

[Kro_Nolius]

Hey Folks,

I installed activepieces on docker and am running ghost in the same place. However, these are behind traefik through a cloudflare tunnel. So lots of pieces to break things.

I’ve been using Zapier, but prefer to have my own thing so am replacing that with activepieces. In Zapier all works fine, as expected.

I have exactly the same problem in both the self-hosted and the cloud activepieces instances.

Workflow:
Create a new flow, choose ghostcms as app, select post published as trigger, select connection, and then “test trigger”
In ghostcms, publish a post and wait for trigger to capture it in activepieces…and crickets.

Exact workflow fails in cloud and self-hosted.

I have confirmed the API URL and the Admin API Key, and they are both correct. Changed them a few times just to make sure. Also, there are no errors when setting up the connection, it goes through fine.

Configs:
docker-compose:

version: '3.0'
services:
  activepieces:
    image: ghcr.io/activepieces/activepieces:0.26.1
    container_name: activepieces
    restart: unless-stopped
    ## Enable the following line if you already use AP_EXECUTION_MODE with SANDBOXED or old activepieces, checking the breaking change documentation for more info.
    ## privileged: true
    ports:
      - 8081:80
    depends_on:
      ap_postgres:
        condition: service_started
      ap_redis:
        condition: service_started
    env_file: .env
    labels:
      traefik.enable: true
      traefik.http.routers.auto.entryPoints: https
      traefik.http.routers.auto.middlewares: auth@file
      traefik.http.routers.auto.rule: Host(`ha.haha,.ha`)
    networks:
      - web

  ap_postgres:
    image: 'postgres:14.4'
    container_name: ap_postgres
    restart: unless-stopped

    environment:
      - 'POSTGRES_DB=${AP_POSTGRES_DATABASE}'
      - 'POSTGRES_PASSWORD=${AP_POSTGRES_PASSWORD}'
      - 'POSTGRES_USER=${AP_POSTGRES_USERNAME}'
    volumes:
      - /opt/appdata/ap_postgres_data:/var/lib/postgresql/data
    networks:
      - web
  
  ap_redis:
    image: 'redis:7.0.7'
    container_name: ap_redis
    restart: unless-stopped
    volumes:
      - /opt/appdata/ap_redis_data:/data
    networks:
      - web

networks:
  web:
    external: true

.env file:

AP_ENGINE_EXECUTABLE_PATH=dist/packages/engine/main.js
AP_API_KEY=Mary had a little lamb, little lamb, little lamb, Mary had a little lamb, it's fleece was white as snow
AP_ENCRYPTION_KEY=Mary was a weird little girl, but we all loved her
AP_JWT_SECRET=Later we cut off her head because she asked us to eat cake instead.
AP_ENVIRONMENT=prod
AP_FRONTEND_URL=https://ha.haha.ha
AP_WEBHOOK_TIMEOUT_SECONDS=30
AP_TRIGGER_DEFAULT_POLL_INTERVAL=5
AP_POSTGRES_DATABASE=activepieces
AP_POSTGRES_HOST=ap_postgres
AP_POSTGRES_PORT=5432
AP_POSTGRES_USERNAME=postgres
AP_POSTGRES_PASSWORD=Little Jack Horner had other ideas for Mary's head though, so we locked him up.
AP_EXECUTION_MODE=UNSANDBOXED
AP_REDIS_HOST=ap_redis
AP_REDIS_PORT=6379
AP_SANDBOX_RUN_TIME_SECONDS=600
AP_TELEMETRY_ENABLED=true
AP_TEMPLATES_SOURCE_URL="https://cloud.activepieces.com/api/v1/flow-templates"

So, here we are.

Help?!

Edit: I also have authelia set to bypass the URI, so there should not be any issues. Other services that bypass are working fine.

[Kro_Nolius]

Anyone finding this later, here’s the issue:

Authelia, was not allowing the bypass. So, I added a resources stanza for the domain as -

• domain:
  - “ha.haha.ha”
  resources:
  - ‘^/ghost/api.$’
  policy: bypass

And it all works now!