Container Manager & RunPod

The Container Manager launches and manages GPU containers on RunPod for AI processing tasks. Different containers provide different capabilities — TTS, video generation, image generation (via ComfyUI), and more.

Container Module

The Container module (container) manages the Container Manager panel.

Module Parameters

Parameter	Type	Description
`visible`	bool	Container Manager panel visibility

Commands

Command ID	Description
`container:toggle-visible`	Toggle the Container Manager panel

pl.cmd("container:toggle-visible")

Container Types

Container	GPU	Purpose	Services
TTS Pod	Any NVIDIA	Text-to-speech, stem splitting, transcription	Qwen TTS, Demucs, Whisper
Video Gen Pod	High VRAM (24GB+)	AI video generation	WAN 2.2, Hunyuan, LTX
ComfyUI Pod	High VRAM (24GB+)	Node-based AI workflows	ComfyUI + custom nodes

Setup

Prerequisites

RunPod Account — Sign up at runpod.io
RunPod API Key — Generate at RunPod dashboard > Settings > API Keys
Configure in Plinken — Enter API key in Settings > Account > RunPod API Key

Launching a Container

Open the Container Manager
Select a container template (TTS, Video Gen, or ComfyUI)
Choose a GPU type
Click Start Pod

The manager handles:

Pod creation via RunPod API
SSH tunnel setup for secure communication
Service health checking
Automatic URL routing to the active pod

GPU Selection

Choose a GPU based on your workload:

GPU	VRAM	Best For
RTX 4090	24 GB	ComfyUI, Video Gen
RTX A6000	48 GB	Large models, batch processing
A100	40/80 GB	Maximum VRAM for largest models
RTX 3090	24 GB	Budget option for Video Gen
RTX 3080	10 GB	TTS only (lower VRAM sufficient)

Pod Lifecycle

State	Description
Creating	Pod is being provisioned on RunPod
Starting	Container is booting and loading models
Running	Services are ready — Plinken connects automatically
Stopping	Pod is shutting down
Stopped	Pod is terminated — no charges

Cost Management

RunPod charges per-GPU-hour while pods are running. Tips:

Stop pods when not in use — The Container Manager makes it easy to start/stop
Use spot instances for lower cost (may be interrupted)
TTS pods need less GPU — use smaller GPUs to save money
Check the RunPod dashboard for current pricing

SSH Tunnel

Plinken establishes an SSH tunnel to the running pod for secure communication. This means:

All traffic is encrypted
No public port exposure needed
Automatic reconnection on network changes

The tunnel routes local URLs to the pod’s internal services.

Active Pod Routing

When a pod is running, Plinken automatically routes requests to the correct service:

Service	Module	URL Property
TTS	`tts`	`tts_url`
Video Gen	`videogen`	`videogen_url`
ComfyUI	`comfyui`	`comfy_url`

Other modules check for active pods before showing generation options:

-- Check if TTS pod is available
local has_tts = plinken.get("tts", "visible")  -- TTS panel adapts to pod state

-- Check if ComfyUI pod is available
local has_comfy = plinken.get("comfyui", "visible")

Multiple Pods

You can run multiple pods simultaneously for different tasks:

TTS pod for voice generation
ComfyUI pod for image workflows
Video Gen pod for video clips

Each pod type is managed independently with its own lifecycle.

Troubleshooting

Issue	Solution
Pod stuck in Creating	Check RunPod dashboard for GPU availability
Connection failed	Verify RunPod API key in Settings
Service not ready	Wait for model loading (can take 2–5 minutes)
High latency	Choose a RunPod region closer to you
Pod disappeared	Spot instance may have been interrupted — restart

Create Overview — AI tools overview
ComfyUI — Node-based AI workflows
WaveSpeed — Cloud video generation (no container needed)