Check LLM Service Connectivity 🐑

CheckLLMServiceConnectivity does one small, extremely useful thing: it pings the LLM service behind a connector and tells you whether it actually works. Before you wire an API key through a whole chain of prompt-generation nodes and burn a render figuring out why nothing came out, you run this and get a straight answer - connected, or not, and why.

It's part of ComfyUI-MieNodes, MieMieeeee's utility pack, in the "LLM Service Config" group, right next to the Set*LLMServiceConnector nodes it's built to test.

Why it earns its spot

The LLM half of this pack has a lot of moving parts before it does anything: the right key, in the right place (node input vs mie_llm_keys.json), under the right config_key, for the right model. Get any of those wrong and the failure shows up downstream, at the prompt generator, as a vague error - which is the worst place to debug it, because you can't tell if the problem is the key, the config, or the prompt node itself. Standard troubleshooting practice is to isolate the failing stage; this node isolates the connector stage for you. Wire your Set* connector straight into this, run, read the result. If it's green, the LLM is fine and any later problem is elsewhere. If it's red, you've found your bug before wasting time.

How it works

Dead simple:

• llm_service_connector (required) - the connector you want to test, from any Set*LLMServiceConnector node (MiniMax, SiliconFlow, ZhiPu, and the rest).

The single output is log (STRING) - the connectivity result and, when it fails, the reason. Pipe it into a ShowAnything node so you can actually read it.

That's the whole node. It doesn't generate anything or change state - it's a diagnostic probe.

Installing it

Install the pack. ComfyUI Manager → search ComfyUI-MieNodes (ComfyUI_MieNodes) → install → restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/MieMieeeee/ComfyUI-MieNodes

then restart. No model download. Nodes appear under the sheep-emoji 🐑 MieNodes menu.

Reading the failures - the actual troubleshooting hub for the LLM nodes

Since this is where you'll diagnose connector problems, here's what a red result usually means:

• Auth failure / invalid key - the key is wrong, expired, or the wrong type for the service. The classic case: a MiniMax sk-cp-... Token Plan key wired into the standard Open Platform connector (or vice versa). Match the key format to the connector.
• Empty key - prefer_local_config is on (the default), so the connector is reading mie_llm_keys.json, but the entry named by config_key is blank or the file isn't there. Fill in the entry, or flip prefer_local_config off to use the node's api_token instead.
• Wrong config_key - you stored the key under one name and the connector's looking for another. Line them up.
• Timeout / network - the service is unreachable from the ComfyUI host, or a firewall is in the way. On a remote GPU this is more common than locally.

Run this the moment an LLM prompt node misbehaves. Nine times out of ten the answer is right here, not in the prompt node.