double-take: Timeout when connecting to AIserver

I don’t really know if this is a bug in Double-Take but I’m hoping it is possible to get some more debugging info.

I run a RaspberryPi 4 with CodeServer.AI as docker image, plain and simple:

  double-take:
    container_name: double-take
    image: skrashevich/double-take
#    image: jakowenko/double-take
    restart: unless-stopped
    volumes:
      - /home/pi/.config/appdata/double-take:/.storage
    ports:
      - 3000:3000

I’m running the latest version: 1.35.5:

Piece of logging:

info: Double Take v1.13.5
info: MQTT: connected
info: MQTT: subscribed to frigate/events, frigate/+/person/snapshot
info: Double Take v1.13.5
info: MQTT: connected
info: MQTT: subscribed to frigate/events, frigate/+/person/snapshot

It times out when trying to connect to the AIserver. In the config:

  aiserver:
    url: http://192.168.21.9:32168
    # number of seconds before the request times out and is aborted
    timeout: 15
    # require opencv to find a face before processing with detector
    opencv_face_required: false
    # only process images from specific cameras, if omitted then all cameras will be processed
    # cameras:
    #   - front-door
    #   - garage

In the same config it does connect to comprefase.

Both double take and aiserver are running on the same docker host. It doesn’t matter if I use the redirected port or the internal IP of the docker container.

On the docker host I see the port 32168 being open.

Config of the aiserver:

  codeproject.ai:
    container_name: codeproject.ai
    image: codeproject/ai-server:rpi64
    restart: unless-stopped
    volumes:
      - ${DOCKERCONFDIR}/codeproject.ai/data:/etc/codeproject/ai
      - ${DOCKERCONFDIR}/codeproject.ai/modules:/app/modules
    ports:
      - 32168:32168

Any thoughts of where to start troubleshooting? I tried running the ping command inside the doubletake container, but it the executable isn’t there (docker exec -it double-take /bin/bash).

ps. in aiserver I installed the face processing module and the object-detection modules too.