Google Find Hub devices in Traccar

What you're sending is an invalid request. You should use the device identifier, not index.

Working in my RPI (traccar server), I copied the keys from an Ubuntu server. All the time I get "curl error (7)" but is working fine. I wonder if 5 mintes (300) is a safe time.
thanks

It probably doesn't make sense to use 5 minutes. Google keeps giving me the same old location if I request too frequently. I think maybe something like once an hour is more reasonable.

hi Anton,
I added other device and now both are on tracking, but most of the time are with me :-) , I´m thinking a way to track on demand, maybe with a "custom command" but to the server to run a script, is there any way to do it? Thanks

Maybe you missed the update, but we already commands.

Hi and thanks for this great project!

I’ve successfully configured the microservice to retrieve the position of 5 Google Find My Device, and everything’s working smoothly with a 1-hour update rate. (including the single position update feature. But adding the attributes to each token is a bit tedious. Is there any other option ?)

In fact, this setup is working even better than the original Find My Device app — I’m consistently getting fresher and more accurate positions via the microservice. Amazing work!

---

• Google Detection
  I’m not sure if Google is actively monitoring or limiting this usage, but so far I haven’t run into any issues with bans or throttling.

• Semantic Location Limitation
  When a token is at a known place like “Home,” the Find My Device backend returns a semantic label instead of GPS coordinates. In these cases, the microservice returns a 404, and Traccar doesn’t report any location.

  Feature Request Idea:
  Add a way to convert semantic locations to predefined GPS coordinates using a mapping file (e.g., semantic_locations.json).
  This would be useful for tracking items left at home or work — a niche but valuable improvement.

• Position History = Game Changer
  Being able to get and retain a position history for the token via Traccar is a major upgrade. This goes far beyond what the official app provides.

• Update Frequency
  Has anyone tried using a higher update frequency (e.g.,5 minutes is too low but every 10–15 minutes)?
  I’m curious about how far we can push the limits without triggering rate limits or bans from Google.

---

• Setup Notes (Manual Install)

One minor thing I noticed: in the manual install instructions, the virtual environment activation was missing. Here's a corrected sequence:

cd /opt/
git clone https://github.com/traccar/google-find-hub-sync.git
cd google-find-hub-sync
python3 -m venv venv
. ./venv/bin/activate   #  This line was missing
pip install -r requirements.txt
python3 main.py

For Docker users, here's an excerpt of my working compose.yaml:

1. Clone the repo into your compose.yaml directory:

git clone https://github.com/traccar/google-find-hub-sync.git

1. Add this to your compose.yaml

find-hub:
  container_name: find-hub
  image: google-find-hub-sync:latest
  build:
    context: ./google-find-hub-sync
    tags:
      - "google-find-hub-sync:latest"
  environment:
    AUTH_TOKEN: "1234567890"
    PORT: 5500
    PUSH_URL: http://traccar:5055  # replace with your actual Traccar service name
  volumes:
    - ./periodic_jobs.json:/app/periodic_jobs.json
    - /etc/localtime:/etc/localtime:ro

3 Build and start with

docker compose build
docker compose up -d

Thanks for feedback.

Feature request for semantic positions you can post on GitHub. I think it's a good idea. Not sure when we'll have time, but we should probably support it.

I figured I'd post some summary information about this for those of us still confused by what is happening under the covers.

• This is a microservice which you grant access to a Google account where devices registered in Google Find My hub are listed. It also connects to your traccar server on the Osmand port (5055), essentially acting like one source IP address with multiple GPS trackers behind it.
• Essentially, it gets position data on trackers from Google, converts that into Osmand format data, and shunts it off to traccar via a TCP connection.
• No devices listed in Google report into traccar until you trigger the microservice to do so, which is done via the curl command where you specify the device ID and the polling period
• When you finish getting everything set up, you may still need to manually add the devices into traccar unless you have allowed database.registerUnknown in your configuration file.
• Also note that the microservice listens on TCP port 5500, not to be confused with the connections it makes to traccar on port 5055

So, even more simply, you're taking position data from trackers in google, translating it to Osmand format, and pushing it into traccar.

Great summary. One thing to add is that you don't need to use curl. You can send commands directly from Traccar.

Thank you for your explanation.
I've already prepared my environment – including access to the Osmand port 5055, creating the two devices on my Traccar server, and building the Docker image for the microservice. The login via Chrome – initiated by main.py – was successful. Maybe I'm just too stupid – but which string in the secrets.json should be the AUTH_TOKEN? The string after "auth", "security_token" or anything else?

AUTH_TOKEN is a key that you set yourself. It has nothing to do with secrets.json.

API_KEY is the key set when starting microservice.py, e.g.:
python microservice.py --auth-token MySecretKey --push-url=http://localhost:5055

It acts like a password and is required for sending HTTP REST commands:
curl -X POST -H "Authorization: Bearer MySecretKey" "http://localhost:5500/devices/DEVICE_ID/position-periodic?interval=300"

Note: curl is not included in python:3.11-slim if you are using docker consider adding it to the Dockerfile or changing to the python:3.11 image . Without exposing the microservice port and using only a backend Docker network, sending commands becomes harder. I understand not installing curl in the Traccar container, as it's more exposed to the internet and adding unnecessary tools can increase the attack surface.

@anton: "You can send commands directly from Traccar."

• can you explain this function?
• how can this triggered?

Note: curl is not included in python:3.11-slim if you are using docker consider adding it to the Dockerfile or changing to the python:3.11 image . >Without exposing the microservice port and using only a backend Docker network, sending commands becomes harder. I understand not installing >curl in the Traccar container, as it's more exposed to the internet and adding unnecessary tools can increase the attack surface.

That's absolutely true. But I don't think it will be necessary to add curl to any of the containers involved. In my environment, both the Traccar container and the microservice container run on the same (virtual) Debian machine. The two containers can reach each other via the IP range 172.17.0.0/16 – and this Docker subnet is also accessible from the Debian VM. You could now use the method described by Anton and send the commands directly from Traccar, or even run curl from the Debian VM.

Just for an initial test, I just ran the following on the Debian VM:

curl -X POST -H "Authorization: Bearer 123123123123123" "http://172.17.*.**:5500/devices/<Device-ID>/position-periodic?interval=300"

And it works perfectly.
Since then, the position has been pushed to Traccar every 5 minutes.
What I don't know yet is how to change the interval or stop the entire process I started with curl... :D Does anyone know?

@rolsch https://www.traccar.org/forums/topic/google-find-hub-devices-in-traccar/#post-112573