I have enabled the option “Include Response Headers and Status” but I get nothing back, like I normally do.
I’m suspecting this error message comes from the node itself and not the other party.
The node doesn’t crash if I run a GET to a newspaper website, or whatever non-api coherent endpoint, which tells me the http-node is somewhat robust to “illegal” characters.
If I remove the auth from the node, the response from the api-endpoint tells me that auth is missing so I guess it has something to do with the auth-process.
I don’t know how to troubleshoot it further. Any ideas?
If you suddenly receive text/html from an API, it might look like an HTML page is being loaded. This often happens when APIs are used excessively (keep an eye on rate limits). In some cases, APIs respond with a default error page in HTML format. This is quite common with APIs that are behind Cloudflare.
For basic HTTP requests, you don’t necessarily need a loop. N8N offers a lot of features for this, including the ability to process HTTP requests in batches and add waiting times between requests. It might be worth trying!
Thanks for your prompt reply!
What I have done in the troubleshooting process is that I removed as many factors as I could. I set up a brand new workflow with a manual trigger and an http-node with a plain GET-request.
Even when I intentionally misspell the url (meaning I get no serious payload) I get this error.
Regarding your suggestion with an HTML page being loaded I wanted to rule that out so I had a newspaper webpage loaded without receiving an error.
Also, I ran the same request against https://webhook.site/
just to see if I sent something weird but it all came back blank.
How do you guys go about troubleshooting the http-node? What bugs me is that the response is not being presented in the $response. It complains about “unsupported content type” but why not display the data to the n8n UI or provide me with some tool to handle the “faulty” data?
If I remove the auth I can access any “shitty” webpage in the world and receive what ever it throws at me. But as soon as I activate the auth I get this error.
For the HTTP node, there are some specific options you can configure. Most of the time, it is sufficient to enable the Continue (using error output) option in the node’s settings.
Thank you for your suggestion, Franz!
I didn’t think of that. Unfortunately it only resulted in me getting the same information (or lack thereof) to the error-leg
It would be great if I could understand what specific part of the data it considers to be unsupported as well as present the data to me so that I can do something about it.
I will not pretend I have the exact same case, but very similar. When I try to use OAuth2 API to setup credentials for Threads, I am able to get the login page to open, to validate the connection from the “Threads” popup. But as a response, the page at the callback opens on the given error : " Error: Unsupported content type: text/html; charset=“utf-8”… Failed to connect".
This error is thrown by teh callback url, while the needed “code” and eventually “state” are available in the url as query parameters.
I don’t know if it can be with any impact but I am behing a reverse proxy with configuration self-hosted 1.83.2/Docker
Thanks for your suggestion. Yes, being behind the reverse proxy might be an issue, depending on if the reverse proxy is doing something weird whilst SSL-stripping the traffic or similar. I am trying different setups in my reverse proxy, thanks to your notice about the reverse proxy.
However, the key issue still remains:
The node does not display the data that it complains about, so we don’t really know what’s wrong.
If we could see the data we wouldn’t have to guess.
I don’t believe being behind a reverse proxy is an issue since the outgoing traffic is not affected by the reverse proxy. Hence, there is probably something in the response that the node doesn’t like and the output doesn’t show me WHAT the node doesn’t like, only that it’s something.
