Webhook help

Questions
,
1

Hi,

First off, i’m not a developer of any kind and English also isn’t my motherlanguage so pardon me for any wrong “things” I describe here but I would greatly appreciate any help.

I’ve got a n8n instance selfhosted on my Raspberry Pi. Everything works fine and I’ve got it routed through a cloudflare tunnel so it even has its own n8n.domain.com for some credentials that don’t work with IP adresses (I’m looking at you Google!).

I’ve got a workflow setup that has a couple of “approval” wait nodes set up: a mail node followed by a webhook wait node. In the mail are some buttons for me to press, like approve or discard, which call upot the wait for webhook call node and they work… sometimes.

After some research I’ve figured out it’s not really a n8n issue, but more of a cloudflare tunnel issue. It takes a while for my call to reach the webhook through the cloudflare tunnel, sometimes it even takes up to over a minute, and cloudflare has a timeout limit on its free tier that blocks the full call if it doesn’t really load. If I refresh the page that is doing the call I get ““code”:409,“message”:“The execution "1234 has finished already.””

Now I’ve been thinking on a way to potentially fix this, but I’m not capable to get it to work: I want my domain to remain as is (n8n.domain.com) to make sure the redirect URI of the credentials I’ve set up to keep working. But other than that I would like the webhooks to still use and listen to my local IP adress (http://84.xxx.xxx.xxx:5678/webhook).

I’ve tried multiple launch parameters. like N8N_HOST=(IP adress) or N8N_LISTEN_ADRESS=0.0.0.0 or WEBHOOK_URL=(IP adress), if I use this last one it uses my IP adress in the link it sends to my mailbox for approval but it keeps listening on https://n8n.domain.com/webhook-waiting/1234 instead of http://IPADRESS/webhook-waiting/1234

Is there a way force the listening for webhooks calls on a specific adress?
Or does anyone have a better way to achieve better webhook wait nodes through a cloudflare tunnel? (I’ve also set the EXECUTIONS_PROCESS to main)

Thanks for any help!

  • **n8n version:1.67.1
  • **Database (default: SQLite): not really applicable
  • **n8n EXECUTIONS_PROCESS setting (default: own, main):main
  • **Running n8n via (Docker, npm, n8n cloud, desktop app): docker
  • **Operating system: Raspberry Pi, don’t know what version
2

Hi @Yorrick_S ! In general most load balancers, proxies or other network components don’t allow long running connections and have a timeout (usually around 1min). Usually how this is handled is to architect around this: The first webhook only accepts the request, stores its details somewhere and responds immediately with the ID that identifies this specific request. Then you have another webhook that can be used to check the status of that work. The approach is described in more detail for example here.

2 Likes
3

Ah I see, I’m gonna try to split up my workflow into multiple seperate workflows that wait for a new trigger.