(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)
Share the output returned by the last node
Hello n8n Community,
I’m experiencing an issue with my WhatsApp Business Cloud API integration.
Workflow: Assistente pessoal WhatsApp
Meta App ID: 1504964524809075
WhatsApp Business Account ID: 1775385397062827
The WhatsApp Trigger is configured for “Messages”, and the workflow is published, but incoming WhatsApp messages are not triggering any executions.
I have confirmed that the Meta app is published, the WhatsApp number is connected, the app is subscribed to the correct WhatsApp Business Account, and the n8n OAuth credential connection test succeeds.
Unpublishing and republishing the workflow did not resolve the problem. The Meta production configuration screen shows empty callback URL and verification token fields.
When I manually execute the trigger, n8n reports that the WhatsApp App ID already has a webhook subscription.
How can I diagnose and restore the production webhook registration without disrupting the existing WhatsApp Business Account subscription?
Thank you!
Hi @andrecury welcome to the N8n community.
The manual-test warning means n8n found an active app-level webhook subscription pointing at a different callback from the test trigger. Your production URL can cause that conflict too. Pause manual tests on this app for now, because n8n’s failed-test cleanup can delete that app-level subscription.
Before deleting anything, use an App Access Token to read GET /<APP_ID>/subscriptions and inspect callback_url, active and fields. Does the callback match this workflow’s production URL, a test URL or another workflow? Please share active, fields, which URL it matches, your n8n version and whether you’re on Cloud or self-hosted. Keep the token private. Since republishing already failed, that readout is needed before choosing a restoration step.
This issue typically happens when Meta’s dashboard loses sync with the webhook URL or when a prior subscription lock prevents n8n from overwriting the callback endpoint. You can resolve the conflict by using the Meta Graph API Explorer to clear the existing subscription and re-establishing the webhook bind directly from the n8n credential setup. Let me know if you need the exact Graph API endpoint path to clear that subscription lock and I can outline the call right here.
Adding the WABA side to the subscriptions check above, since n8n’s WhatsApp Trigger docs cover this in Common issues. The “already has a webhook subscription” message is about the app’s webhook URL, not your WABA, so don’t send a DELETE to /<WABA_ID>/subscribed_apps to clear it. That unsubscribes the app and Meta stops sending anything. Do run a GET on /<WABA_ID>/subscribed_apps though. If your app’s entry has an override_callback_uri, Meta sends message webhooks for that WABA there instead of to your n8n URL, and an override set on the phone number wins over both. The same page says Meta keeps one webhook per app, so a manual test run and the published workflow can’t both receive.
One addition to the subscription checks above: before changing anything, record all three places Meta can redirect this event.
GET /<APP_ID>/subscriptions with an app access token: capture callback_url, active, and fields.
GET /<WABA_ID>/subscribed_apps: check whether this app is listed and whether an override_callback_uri is present.
Check for a phone-number-level callback override in Meta, if one exists.
Compare every callback URL against this workflow’s production URL, not the /webhook-test/ URL. A manual trigger run can move the app-level callback to the test URL, and a WABA or phone-number override can win even when the app subscription looks correct.
After making one change, stop manual tests, republish once, and send from a non-test number. Require both proofs: an entry in Meta’s webhook delivery log with HTTP 200, and a new n8n execution containing the message payload. If Meta has no delivery attempt, it is a subscription/routing problem. If Meta receives 200 but n8n has no execution, check n8n ingress and activation. If an execution exists but later nodes fail, the trigger is already working.
Please capture those values before deleting any subscription.
The subscription and override checks above cover the likely causes. Before changing registration again, could you share a small sanitized matrix: production callback matches (yes/no), messages field subscribed (yes/no), WABA override present (yes/no), phone-level override present (yes/no), and the HTTP status/timestamp of the latest real inbound delivery? Also include your n8n version and Cloud vs self-hosted. Please omit tokens and private message content. A successful OAuth credential test proves access, not that inbound events reach the production callback. That matrix should make it easier to choose one targeted fix rather than deleting or recreating subscriptions speculatively.