Home Assistant addon for OCW

[Chaoscontrol]

Here we go.

Not fully tested yet. But it runs, it sends SMS. I still need to update my URL in the car to test properly that it receives data. But otherwise, it seems to work!

[Chaoscontrol]

Is there any trick to adding a second connection to the Navi? I have added it using the VFlash editor, as shown here. I kept all values the same, just changed the server URL (to use my private HA one) and the connection name to distinguish it.

I wrote the VFlash in the car fine, but then when opening the provider list in the Navi, only 1 connection shows up (the first one). The name was updated, which confirms the VFlash restore worked fine. But the second connection is not there.

Any idea why?

[developerfromjokela]

HTTP port is what navi unit needs, port 55230 is used for remote commands with TCU, no usage from navi. Port for TCU also can't be configured, it's hard coded as 55230

[Chaoscontrol]

I'm sorry but I still don't understand. Is the HTTP port the navi needs the same as the webUI port? Or a different one?

I have now used: http://ocw.mydomain.com/WARCondelivbas/it-m_gw10/ in the Navi, and it doesn't work either.

I have also tested the Android app, using custom server option. Used my webUI domain. It refreshes the page as if it was successful (it even loads the changelog popup), but reloads the login page without no error message. I even can see a log in my private server: HTTP POST /api/token/obtain/ 200 [0.59, 172.30.32.1:34978], which means it's receiving the request fine.

[developerfromjokela]

I'm sorry but I still don't understand. Is the HTTP port the navi needs the same as the webUI port? Or a different one?

I have now used: http://ocw.mydomain.com/WARCondelivbas/it-m_gw10/ in the Navi, and it doesn't work either.

I have also tested the Android app, using custom server option. Used my webUI domain. It refreshes the page as if it was successful (it even loads the changelog popup), but reloads the login page without no error message. I even can see a log in my private server: HTTP POST /api/token/obtain/ 200 [0.59, 172.30.32.1:34978], which means it's receiving the request fine.

HTTP Port for GUI is also used for Carwings connection between server and navi. So yes, same port. If the OCW server HTTP port is not exposed to internet as 80, you'll have to set it in URL to correct port.

[Chaoscontrol]

What should happen when accessing http://biz.viaaq.eu/WARCondelivbas/it-m_gw10/ ?

That path in my server is currently not accessible, and I'm trying to fix that. When accessing that URL it gets redirected to http://biz.viaaq.eu/, with a basic page with instructions about the gateway. Is that what's expected?

[developerfromjokela]

What should happen when accessing http://biz.viaaq.eu/WARCondelivbas/it-m_gw10/ ?

That path in my server is currently not accessible, and I'm trying to fix that. When accessing that URL it gets redirected to http://biz.viaaq.eu/, with a basic page with instructions about the gateway. Is that what's expected?

Yes. GET and POST requests that don't come from the car redirects to the main page.

[Chaoscontrol]

What method are you using in your public instance server to expose ports HTTP 80 and TCP 55320 to the internet?

I have been trying to do this from the addon in several ways, but it's not doable so far.

• I've tried using Cloudflare as reverse proxy. It works for HTTP 80, but it doesn't allow exposing custom TCP ports.
• I've integrated Tailscale Funnel into the addon, and then found out that TCP only works for a few specific ports, and doesn't include 55230 either.
• AI suggests my only options is having a public VPS to forward the ports (which defeats the purpose as it would be basically the same as your public instance) or just make the user forward their ports in their router. I was trying to avoid this to make config easy for the users, and for better security. But right now I don't know any other options.

Any ideas?

[Chaoscontrol]

So do you think FRP doesn't require a VPS?

I already tried going that route and chatGPT told me this.

Is it wrong? Or maybe I just don't understand enough.

---

So: you’ve installed only the FRP add-on in Home Assistant, and there is no external server (VPS) — correct?

Then here’s the key point:

❗FRP always needs two components — a server (frps) and a client (frpc).
The Home Assistant add-on is the client (frpc). It can’t expose ports to the public internet on its own — it only connects out to a remote frps.

That’s why I mentioned the VPS earlier: you need somewhere for frps (the “listener”) to live.
Without it, FRP in HA doesn’t know where to forward the traffic.

---

🧩 Think of it like this:

• frpc (HA add-on) → “Hey, I have port 55230 open locally.”

• frps (server) → “Okay, I’ll expose that to the world as publicip:55230.”

So unless Home Assistant can directly open public TCP ports itself (which it can’t, because it’s behind NAT, Cloudflare, etc.), you need that middleman somewhere publicly reachable — even a free-tier cloud VM, another NAS with public IP, or a small always-on Raspberry Pi at a location with a routable IP.

---

If you really don’t want to run a VPS, there are a few alternatives that achieve the same goal without hosting anything yourself:

Option	What it does	Notes
Ngrok / LocalTunnel / Cloudflare Tunnel (HTTP only)	Expose internal ports to internet	But none support raw TCP in your case
Tailscale Funnel (Beta)	Exposes TCP from inside your network	Would require Tailscale CLI, not the HA add-on
Expose via your router	Port-forward 55230 directly to HA	Works, but you’ll need static IP or DDNS, and Cloudflare won’t help here

---

So to recap:

• ✅ The HA FRP add-on alone = client only

• ❌ It cannot publicly expose ports by itself

• 🧠 You need an FRP server (frps) somewhere public to “relay” connections

---

Would you like me to help you pick the easiest free/cheap place to host the small frps server? (it’s very lightweight — runs even on free Oracle Cloud, Fly.io, or a $5 VPS).

[developerfromjokela]

You do need one low-power host in vps or somewhere accessible from public internet. frp does not need huge computational power, you could use something like oracle free tier to run it.

Without port forward or other kind of tunneling to port 55230 it won't work :( the port is hard-coded in TCU firmware

[Chaoscontrol]

Ok, so you agree it needs a VPS then. I thought the server needed to be run on the TCU though, not on the VPS. That's kinda doable then.

I'll check the Oracle Free Tier, although not a fan of providing an addon with that "constraint". As if that VPS disappears, the addon won't work anymore. It'll be fragile.

[developerfromjokela]

Ok, so you agree it needs a VPS then. I thought the server needed to be run on the TCU though, not on the VPS. That's kinda doable then.

I'll check the Oracle Free Tier, although not a fan of providing an addon with that "constraint". As if that VPS disappears, the addon won't work anymore. It'll be fragile.

Other solution is to battle the ISP to get port forward 😄

[Chaoscontrol]

I have managed to set up a VPS and expose both ports via FRP successfully. However, I have realised there needs to be something additional done to allow this VPS to handle multi-user (everyone that will install the addon).

Do you have an idea of how to resolve this? I would need either each user assigned a unique subdomain, and/or a unique auth token. And I'm not sure frps.ini allows "infinite" auth tokens, or I have to define a list of them, which would be problematic.

For the subdomain, I was trying to do this with duckdns to avoid exposing a personal domain of mine. But duckdns doesn't allow me to create wildcard subsubdomains. Not sure of alternatives.

[Chaoscontrol]

@developerfromjokela I've continued working on this and I have a couple questions.

I've noticed that OCW uses 2 external services for the charging stations on the navi. OpenChargeMap and Iternio.

• As far as I was able to find, Iternio API key is paid. Is this right? Did you pay for it? What's the cost? I couldn't see it in the website as they only say to send an email.
• For both keys, in the public server, all users use your own keys? Is there no cap for requests?

I'm implementing them as optional in the addon.

Thanks.

[developerfromjokela]

@developerfromjokela I've continued working on this and I have a couple questions.

I've noticed that OCW uses 2 external services for the charging stations on the navi. OpenChargeMap and Iternio.

• As far as I was able to find, Iternio API key is paid. Is this right? Did you pay for it? What's the cost? I couldn't see it in the website as they only say to send an email.

• For both keys, in the public server, all users use your own keys? Is there no cap for requests?

I'm implementing them as optional in the addon.

Thanks.

Hi.

Openchargemap api key is free from their website. The key is common for whole server so each user in the server does not have its own openchargemap key.

[Chaoscontrol]

Are you sure about that? Maybe it changed?

What about the Iternio one?

[developerfromjokela]

Are you sure about that? Maybe it changed?

What about the Iternio one?

You register to openchargemap and your api key is under account settings. You just insert it into the config of opencarwings server.

For iternio you have to contact them, but I have been considering switching away since it's hard to get

[Chaoscontrol]

That was my question. Yes, OpenChargeMap is free, but every user will need its own. My question was, for the public server, everyone is using yours, right?

And iternio says to contact them, but it also says it has a price. Do you know how much? I don't think I'll be getting one of those.