Challenges Implementing A Reliable Queueing Mechanism

Questions
,
1

Why Am I Here ?

I’m building a Telegram virtual assistant using n8n to automate responses to user queries. The bot should handle both text and audio messages, leveraging AI for intelligent responses.

Current Progress…

Workflow Foundation: I’ve set up the basic n8n workflow with a Telegram Trigger node, Switch node (to separate text and audio), Edit Fields node, AI Agent node, and Telegram Output node.

Audio Transcription: Integrated OpenAI’s transcription tool to convert audio messages to text before sending them to the AI Agent.

AI Integration: Connected the AI Agent node to Anthropic and Google for generating responses.

Challenges…

Queueing Mechanism: I’m facing challenges implementing a reliable queueing mechanism to handle user messages sequentially. I have two options:

  1. Code Node Approach: Using code nodes for queueing leads to “Unknown error” messages that need debugging.

  2. Redis Approach: Implementing Redis requires assistance with installation, configuration, and integration with n8n.

Seeking Assistance On !

I’m looking for a JavaScript developer or someone with Redis expertise to help me:

Option 1 (Code Node)…

  • Debug and resolve the “Unknown error” messages in the code nodes.
  • Ensure smooth data flow through the workflow using the code-based queueing mechanism.

Option 2 (Redis)…

  • Install and configure Redis on my server or guide me through using a managed Redis service.
  • Integrate Redis with n8n, including setting up the necessary nodes and configuring them correctly.
  • Migrate the existing queueing logic to utilize Redis for improved reliability and performance.

I am also open to other suggestions…

2

Hey Adams, what’s up?

Saw you’re stuck on the queueing mechanism for your Telegram automation with n8n. Cool setup, by the way—really like how you’re integrating Anthropic and Google.

For a quick fix, definitely review your Code Node logic. Just throw in some simple try/catch blocks and add console.log() statements to pinpoint exactly where those “unknown errors” are coming from. Usually sorts it out fast.

Redis is a solid step up for a more robust solution without too much hassle. Spin it up quickly via Docker (docker run -d redis) or grab a free instance on Redis Cloud. Then just integrate using n8n’s HTTP Request node with commands like LPUSH (enqueue) and BRPOP (dequeue). If you need some ready-to-use examples, just ask.

There are even beefier solutions like Kafka, Kubernetes, or going fully serverless with AWS Lambda—great if your project scales up significantly. But unfortunately, I can’t detail too much here or else I might get mistaken for AI again, haha!

If you wanna chat more deeply about this or need real hands-on help with Redis or any other option, just click the “Message” button (assuming it really is private, lol) and I’ll be happy to help.

Cheers,
Dandy

3

Hey Adam,
i have a question on this part of your workflow.

I’m trying to solve this one, I’m getting a webm or a audio ogg file and neet to convert it to something that open ai will accept.
since you’ve solved it can you help me here.
id appreciate any help, thx