Clarify limitation MCP trigger limitations in queue mode (#3084)

This commit is contained in:
oleg 2025-04-15 18:34:12 +02:00 committed by GitHub
parent 18118a7e28
commit ca608e5a3a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -94,7 +94,16 @@ Be sure to replace the `<MCP_URL>` and `<MCP_BEARER_TOKEN>` placeholders with th
## Limitations
The MCP Server Trigger node doesn't yet support n8n instances running in [queue mode](/hosting/scaling/queue-mode.md) mode.
**MCP Trigger and Webhook Replicas**
The MCP Trigger node relies on Server-Sent Events (SSE), which require the same server instance to handle persistent connections.
* If you run n8n in [queue mode](/hosting/scaling/queue-mode.md) with only a **single webhook replica**, MCP Trigger will work as expected.
* If you run **multiple webhook replicas**, make sure that you route all `/mcp*` requests to a single, dedicated webhook replica.
* To achieve this, create a separate replica set with one webhook container for MCP requests, and update your ingress or load balancer configuration to direct all `/mcp*` traffic to that instance.
/// warning | MCP Trigger with Multiple Webhook Replicas
If you run MCP Trigger with multiple webhook replicas and don't route all `/mcp*` requests to a single, dedicated webhook replica, SSE connections will frequently break or fail to deliver events reliably.
///
## Related resources