n8n-nodes-resilient-sse

This is an n8n community node. It lets you start workflows from a Server-Sent Events (SSE) stream.

Its focus is staying connected. n8n's built-in SSE Trigger reconnects on a fixed 3-second interval and gives up permanently when the server returns an HTTP error. This node reconnects with exponential backoff and jitter, resumes with Last-Event-ID, honours the stream's own retry: hint, and surfaces the connection state so you can see what it is doing.

n8n is a fair-code licensed workflow automation platform.

Installation
Operations
Credentials
Compatibility
Usage
Resources
Version history

Installation

Follow the installation guide in the n8n community nodes documentation.

To install an unpublished local build instead, build a tarball with npm pack and install it into your n8n nodes directory:

mkdir -p ~/.n8n/nodes && cd ~/.n8n/nodes
npm install --omit=peer /path/to/n8n-nodes-resilient-sse-1.0.1.tgz

Then restart n8n. --omit=peer matters: without it npm tries to install the n8n-workflow peer dependency, which pulls in isolated-vm and fails compiling native code. n8n provides n8n-workflow itself.

Operations

The node connects to an SSE endpoint and emits one workflow item per received event.

Event payloads are parsed as JSON and emitted as the item's JSON. A payload that is not valid JSON — a plain-text heartbeat, a [DONE] sentinel — is emitted as { "data": "<raw text>" } instead, so an unexpected frame never tears the stream down. Arrays and scalars are wrapped the same way, since spreading them would produce numeric keys.

With Include Metadata on (the default), each item also carries a $metadata key:

{
  "seq": 9,
  "$metadata": {
    "eventType": "update",
    "lastEventId": "evt-9",
    "origin": "http://example.com/events",
    "timestamp": "2026-09-10T12:00:00.000Z"
  }
}

eventType is the stream's event: field, or message when it is absent — this is how you tell named events apart.

Credentials

Two optional credential types, selected by the node's Authentication parameter:

Both store their secret encrypted at rest, which is why they are preferable to putting a token into a custom header in the node parameters.

Neither carries a service URL of its own — the endpoint lives on the node — so there is nothing to send a probe request to. The Test button therefore only confirms the fields are filled in; the credential is genuinely proven the first time the trigger connects.

Choose None for unauthenticated endpoints.

Compatibility

Requires an n8n version whose runtime provides the global fetch API (Node.js 18+). The node has no runtime npm dependencies.

Usage

Parameters

Three parameters are always visible:

Everything else lives under Additional Fields, and the defaults below apply whether or not you add the field:

Retry delays are floored at 50 ms, so a zero or malformed delay cannot turn reconnecting into a hot loop.

How reconnecting works

A reconnect is triggered by a retryable HTTP error, a mid-stream connection drop, a connection timeout, or the server closing the stream cleanly. The delay is Initial Retry Delay × Backoff Factor ^ (attempt - 1), capped at Max Retry Delay, then optionally jittered.

The node tracks the stream's id: field and sends it back as the Last-Event-ID header on every reconnect, so a server that supports replay can resume where the workflow left off.

Client errors are not retried. A 4xx response other than 408 Request Timeout or 429 Too Many Requests means the request itself was refused, so repeating it unchanged cannot succeed. The node stops and reports a fatal error rather than hammering the endpoint — check the URL, the credential, and any custom headers.

When Max Retry Attempts is exceeded, the node likewise reports a fatal error to n8n, which deactivates the workflow and retries activation with its own backoff.

Seeing the connection status

Every state change (connecting, connected, disconnected, error, reconnecting, stopped) is written to the n8n logs.

Enable Emit Connection Status to receive the same transitions as workflow items, which is the easiest way to see them in the editor:

{
  "status": "reconnecting",
  "url": "http://example.com/events",
  "attempt": 2,
  "message": "Stream closed by the server",
  "timestamp": "2026-09-10T12:00:00.000Z"
}

Status items are emitted on the same output as data items, so branch on the status key with an If or Switch node if you need to separate them.

Be aware of the cost: n8n starts a separate workflow execution per emitted item, so a single connection flap produces three executions (error, reconnecting, connected). Against a flapping endpoint this fills your execution history quickly. Leave it off in production and turn it on when diagnosing a connection problem.

Resources

• n8n community nodes documentation
• SSE specification (WHATWG)

Version history

1.0.1

Fixed the node icon, which was previously invisible against both the light and dark themes.

1.0.0

Initial release.