Request Failed with status 504 - Gateway Timeout

Question

Request Failed with status 504 - Gateway Timeout

Opened this issue 8 months ago · 17 comments

Following the last issue solved lately, I am now having this error back :

error: Error Polling Data: {"error":{"error":{"message":"Request Failed with status 504 - Gateway Timeout","request":{"method":"POST"},"response":{"

Best

Answer 1 · 2024-03-25T13:59:53.000Z

Unfortunately, as noted in https://github.com/BigThunderSR/homeassistant-addons-onstar2mqtt?tab=readme-ov-file#helpful-usage-notes, there's not much we can do about that.

Answer 2 · 2024-03-25T20:23:04.000Z

I'm having a similar issue and the addon fail to start, the reason in the logs is:

error: Main function error: {"error":{"message":"Request Failed with status 503 - Service Unavailable" ... or "message":"No response"

When doing the POST /api/v1/oauth/token. Will it be for the same reason that causes this 504 or maybe it will be something else?

Thanks!

Answer 3 · 2024-03-25T21:32:17.000Z

I'm having a similar issue and the addon fail to start, the reason in the logs is:

error: Main function error: {"error":{"message":"Request Failed with status 503 - Service Unavailable" ... or "message":"No response"

When doing the POST /api/v1/oauth/token. Will it be for the same reason that causes this 504 or maybe it will be something else?

Thanks!

Have you ever been able to use this add-on before? If not, are you based in the US or Canada?

Answer 4 · 2024-03-26T20:21:57.000Z

I don't know if this is related, but I got a weird error response to trying to start the car

{"data":{"commandResponse":{"body":{"error":{"code":"ONS-220","description":"Failed to execute vehicle request.","subCode":null,"subCodeDescription":null}}

After trying the command again it worked just fine. Why don't we retry the command if we got an error back from the API server?

Answer 5 · 2024-03-26T21:51:37.000Z

I don't know if this is related, but I got a weird error response to trying to start the car

{"data":{"commandResponse":{"body":{"error":{"code":"ONS-220","description":"Failed to execute vehicle request.","subCode":null,"subCodeDescription":null}}

After trying the command again it worked just fine. Why don't we retry the command if we got an error back from the API server?

That might be something unique to your Lyric because I have never seen that one before.

In any case, they have so many error codes that it's not a good idea to pound the API automatically whenever an error is returned because that's a surefire way to expedite getting the API blocked forever for all of us...

Answer 6 · 2024-03-26T23:21:01.000Z

The problem is that we have no idea if the command succeeded unless we go in the addon logs. I suppose we can create an automation and subscribe to errors with mqtt. Has anyone done this already?

Answer 7 · 2024-03-26T23:36:42.000Z

The problem is that we have no idea if the command succeeded unless we go in the addon logs. I suppose we can create an automation and subscribe to errors with mqtt. Has anyone done this already?

That's exactly why I added the command response status into MQTT topics for each command.

I personally use my Node-RED project for issuing commands because it's a lot more reliable for the purpose of running commands.

Answer 8 · 2024-03-29T06:32:29.000Z

@sam2kb, I added auto discovery to the command status and related timestamp (triggered whenever a command is run, so won't show anything until you run a command) in V1.15.0, so it should be pretty straightforward to get command status from HA through the MQTT devices item in Settings -> Devices in HA.

Answer 9 · 2024-03-29T16:26:43.000Z

This is awesome. Why isn't it a part of existing mqtt device though? Unless you are planning on adding all supported commands to the list

Answer 10 · 2024-03-29T16:44:31.000Z

This is awesome. Why isn't it a part of existing mqtt device though? Unless you are planning on adding all supported commands to the list

Pretty much that, but also the following.
I tried that first, but I didn't like how it made everything look unclean with non-numeric sensors mixed in with all the numeric sensors. You can combine them as needed in any Lovelace dashboard, so it doesn't affect overall usability in the end.

Answer 11 · 2024-03-29T16:52:11.000Z

You can also shorten the status messages to completed and failed or similar

Answer 12 · 2024-03-29T16:56:30.000Z

You can also shorten the status messages to completed and failed or similar

I could, but I've learned that TMI is better than folks having to dig through things to figure out what any short name refers to among other things. 🙂

Answer 13 · 2024-04-01T04:25:50.000Z

@sam2kb, check out the v1.17.0 release where I added an option to allow you to group everything under one MQTT device if that's what you want to do.

Answer 14 · 2024-04-02T02:02:51.000Z

You are the best!
This add-on and a Home Assistant app in the car OS is what I use all the time. The GMC app is garbage, I hope they won't restrict their API.

Answer 15 · 2024-04-02T02:07:30.000Z

You are the best!
This add-on and a Home Assistant app in the car OS is what I use all the time. The GMC app is garbage, I hope they won't restrict their API.

Unfortunately, I think it's only a matter of time before they decide to completely block us, so I guess let's enjoy it while we can.

Answer 16 · 2024-04-04T15:44:02.000Z

@BigThunderSR sorry in delay for the response!

Have you ever been able to use this add-on before? If not, are you based in the US or Canada?

No and i'm not from US/Canada.. i'm from Argentina (south america).. this api is "geo fenced" in some way?

Note: I have a Cruze and use the android app "myChevrolet" so I assumed that this api works for me.
And I tried to emulate the curl to get the oauth token from the logs buy have no luck (or time/effort) infering the payload because it seems to log the payload in an array of bytes or something like that (i'm not very proeficient with node and its types :P)

Thanks!

Answer 17 · 2024-04-04T16:52:54.000Z

@BigThunderSR sorry in delay for the response!

Have you ever been able to use this add-on before? If not, are you based in the US or Canada?

No and i'm not from US/Canada.. i'm from Argentina (south america).. this api is "geo fenced" in some way?

Note: I have a Cruze and use the android app "myChevrolet" so I assumed that this api works for me. And I tried to emulate the curl to get the oauth token from the logs buy have no luck (or time/effort) infering the payload because it seems to log the payload in an array of bytes or something like that (i'm not very proeficient with node and its types :P)

Thanks!

@axcoro, as far as we know, OnStarJS only works with US and Canadian accounts. You would need to open a request there to see if accounts in other countries can be supported, but I don't think it has been possible so far.