Node Restart on Incorrect Response (Polling)

The API responds successfully (status 200), but the task is still processing. With Retry if This Node Returns Incorrect Data enabled, the node restarts after a set delay while the response contains "trigger" words (e.g. queued, processing). Once the response no longer contains them, the node completes and passes the result forward.

This is especially useful for content generation — video, audio, images, documents.

When to Use This

• Working with "Get Result" type nodes: one node starts the process and returns an ID, another retrieves the result later
• Applicable to almost all AI services (except text-based ones)

Typical example — video generation API:

// First request — task queued
{"status": "queued", "id": "abc123"}

// After 5 seconds — processing
{"status": "processing", "id": "abc123"}

// After 10 seconds — done!
{"status": "completed", "url": "https://..."}

You need to wait for completed with a ready file link. You specify trigger words (queued, processing) — while they're present in the response, the node restarts after the specified time. Once completed arrives (without trigger words), the node completes successfully.

Enable: Retry if This Node Returns Incorrect Data

Configuration

Retry if This Node Returns Incorrect Data — enable automatic retry when the response doesn't match the expected pattern.

Number of Retry Attempts and Delay between Retries (sec) — how often to poll the API (see Node Restart on Error).

REGEXP (Regular Expression) — you specify trigger words: if they're found in the response, the node restarts after the specified time. If no trigger words are found — success, the node completes and passes the result forward.

Important: Without .*, the pattern only matches if the entire response exactly equals your word. Always write .*(your_pattern).*

Pattern Examples

Example: Waiting for Video Generation

Task: A video generation API first returns in_progress, then queued, and finally completed with a ready link.

Configuration:

1. Enable Retry if This Node Returns Incorrect Data
2. REGEXP: .*(in_progress|queued).*
3. Number of Retry Attempts: 5
4. Delay between Retries (sec): 30

How it works:

• Iteration 1: "status": "in_progress" — contains trigger word. Node restarts in 30 seconds.

• Iteration 2: "status": "queued" — contains trigger word. Node restarts in 30 seconds.

• Iteration 3: "status": "completed" — no trigger words. Node completes, result (video) is passed to the next node.

The node restarts every 30 seconds until it receives a response without trigger words. Once a successful response arrives, data is passed to the scenario.

Technical Details

• Uses Go (RE2) regular expression engine
• Supported: \d, \s, \w, |, ()
• Not supported: lookahead (?=...) and lookbehind (?<=...)

What's Next