Sulla Desktop -- Port Reference

All ports used by Sulla Desktop services. Host-side ports in the 301xx range are forwarded into the Lima VM; lower ports run directly on the Electron host.

Port Map

Port	Service	Protocol	Bind Address	Purpose	Source File
3000	ChatCompletionsServer	HTTP	0.0.0.0	OpenAI-compatible API (/v1/chat/completions, /v1/models, /v1/embeddings, /v1/moderations, /v1/integrations)	pkg/rancher-desktop/main/chatCompletionsServer.ts
6107	HttpCommandServer	HTTP	127.0.0.1	Internal CLI/settings server -- settings CRUD, diagnostics, extensions, snapshots, backend state, port forwarding	pkg/rancher-desktop/main/commandServer/httpCommandServer.ts
6108	Terminal WebSocket Server	WS	127.0.0.1	PTY sessions inside the Lima VM via limactl shell 0; xterm frontend connects here	pkg/rancher-desktop/main/terminalServer.ts
6120	Dashboard Server	HTTP/WS	127.0.0.1	Rancher Dashboard UI and K8s API proxy (proxies /k8s, /api, /apis, /v1, /v3 to 127.0.0.1:9443)	pkg/rancher-desktop/main/dashboardServer/index.ts
30114	llama-server	HTTP	127.0.0.1	Local LLM inference (llama.cpp); serves OpenAI-compatible /v1/embeddings and completions endpoints	pkg/rancher-desktop/agent/services/LlamaCppService.ts
30116	PostgreSQL	TCP	127.0.0.1	Persistent relational store (maps host 30116 to container 5432); used by n8n and general Sulla data	pkg/rancher-desktop/assets/sulla-docker-compose.yaml (service: postgres)
30117	Redis	TCP	127.0.0.1	In-memory cache and queue backend (maps host 30117 to container 6379); used by n8n Bull queue	pkg/rancher-desktop/assets/sulla-docker-compose.yaml (service: redis)
30118	WebSocket Hub	WS	127.0.0.1	Internal pub/sub communication bus (maps host 30118 to container 8080); used by ChatCompletionsServer tasker channel	pkg/rancher-desktop/assets/sulla-docker-compose.yaml (service: ws-server)
30119	n8n Workflow Engine	HTTP	127.0.0.1	Workflow automation UI and webhook runner (maps host 30119 to container 5678); backed by PostgreSQL and Redis	pkg/rancher-desktop/assets/sulla-docker-compose.yaml (service: n8n)

Details

3000 -- ChatCompletionsServer

Express server exposing an OpenAI-compatible REST API. Accepts chat completion, completion, embedding, and moderation requests. Also provides an integration API for calling external service endpoints. Listens on all interfaces (0.0.0.0) so it is reachable from the Lima guest and from LAN clients.

Connects to the WebSocket Hub (port 30118) on channel tasker to receive and respond to asynchronous user messages (e.g. Slack app mentions).

6107 -- HttpCommandServer

Authenticated HTTP API used by the rdctl CLI and the Electron renderer to read/write application settings, run diagnostic checks, manage extensions and snapshots, trigger factory resets, and control the backend lifecycle. Credentials are written to ~/Library/Application Support/rancher-desktop/rd-engine.json.

6108 -- Terminal WebSocket Server

Spawns node-pty processes that run limactl shell 0 to provide interactive terminal sessions inside the guest VM. Supports multiple concurrent sessions, automatic retry while the VM is booting, and terminal resize events.

6120 -- Dashboard Server

Serves the Rancher Dashboard static assets and proxies Kubernetes API requests (/k8s, /api, /apis, /v1, /v3, etc.) to the cluster API at https://127.0.0.1:9443. Supports WebSocket upgrades for real-time cluster communication.

30114 -- llama-server

Local llama.cpp inference server. Downloaded and managed by LlamaCppService. Provides embedding and completion endpoints consumed by the ChatCompletionsServer for on-device AI capabilities.

30116 -- PostgreSQL

postgres:16-alpine container. Default credentials: user sulla, password sulla_dev_password, database sulla. Data persisted at /var/lib/sulla/postgres inside the VM.

30117 -- Redis

redis:7-alpine container with AOF persistence (--appendonly yes). Data stored at /var/lib/sulla/redis. Used as the Bull queue backend for n8n job execution.

30118 -- WebSocket Hub

Custom WebSocket relay (byrdziak/sulla-websocket:latest). Provides a channel-based pub/sub bus for inter-service messaging. The ChatCompletionsServer subscribes to the tasker channel to bridge external events (e.g. Slack) into the agent graph.

30119 -- n8n Workflow Engine

n8nio/n8n:latest container. UI security is disabled for local development. Uses PostgreSQL (port 30116) for persistence and Redis (port 30117) for the execution queue. REST API and MCP feature flag are enabled.