Labnana Get Task

This is the other half of the async workflow in ComfyUI-Labnana. Where Labnana Submit Task throws a job over the wall and walks away, LabnanaGetTask goes and fetches it. Give it a task_id and it will either grab the current state immediately or - the part that's actually useful - block and wait for the job to finish, then download the images and hand you a normal ComfyUI IMAGE tensor. The generation keeps cooking on Labnana's servers in the meantime, so you can submit a batch, close ComfyUI, come back tomorrow, and pull the results down.

How it works

It calls the task endpoint for that id, polls at your chosen interval until the status flips to success (or fail), then downloads the result images. Two inputs do the real work:

• wait_for_completion (default true) - when on, the node polls until the job is done, up to timeout seconds. When off, it does a single fetch and errors if the task isn't finished - useful for a quick "is it done yet?" status check.
• poll_interval (default 5s) and timeout (default 600s) - how often to poll and how long to wait before giving up.

task_id is the string you got from Submit Task, or from the task_id output of the main Labnana Image Generation node. The node always re-queries the server on every run, so a stale cached result is never the issue - the server state is.

Outputs: images (the IMAGE batch - wire into Save Image or keep processing locally), status (STRING, e.g. success), image_urls (the public result URLs, valid for 7 days), and info_json (the full task record, pretty-printed).

One honest warning: if you use it with wait_for_completion on and a 4K job is slow, you'll be waiting - raise timeout for big jobs, the default 600s is snug for 4K at busy times.

Install

Same story as every node in the pack - search Labnana in ComfyUI Manager, or:

cd ComfyUI/custom_nodes
git clone https://github.com/exoticknight/ComfyUI-Labnana.git

Restart ComfyUI. Only dependency is requests, and the Labnana API Client node needs your ls_... key (or the LABNANA_API_KEY environment variable).

Gotchas

• A failed job errors when you fetch it - you get the server's failMsg, so the diagnosis is usually right there.
• wait_for_completion off means "fetch once". If the job is mid-flight you'll get an error telling you it's not finished and to turn waiting on. That's by design, not a bug.
• Pair it with the right node. Submit Task → Get Task is the natural pairing for batch pipelines. For a single generation you never need either - Image Generation does submit, wait, and download in one node.

If you only ever run interactive generations, you can skip this entire Advanced menu. The moment you're scripting batch work or fetching results after a restart, though, this node is what makes the fire-and-forget pattern actually pleasant.