---
# Code generated by go run ./scripts/gendocs. DO NOT EDIT.
title: "vllm-helper recipes install-service"
description: "Install a systemd service that keeps the active recipe serving"
---

<!-- DO NOT EDIT | GENERATED CONTENT -->

Installs a unit that runs 'vllm-helper restart' at boot and 'vllm-helper stop' on
shutdown. The unit names no recipe: which one runs comes from the active profile,
so switching recipes does not mean editing the unit. Linux hosts only.

## Usage

```console
vllm-helper recipes install-service [flags]
```

## Options

### --name

|         |                        |
| ------- | ---------------------- |
| Type    | `string`               |
| Default | `vllm-cluster.service` |

systemd unit name

### --now

|      |        |
| ---- | ------ |
| Type | `bool` |

enable and start the service immediately

### --supersede-native

|         |        |
| ------- | ------ |
| Type    | `bool` |
| Default | `true` |

stop and disable the single-node unit this replaces

### --timeout

|         |        |
| ------- | ------ |
| Type    | `int`  |
| Default | `1800` |

seconds to allow for the model to load

### --user

|      |          |
| ---- | -------- |
| Type | `string` |

user the launcher runs as (default: $USER)

### --watchdog

|         |        |
| ------- | ------ |
| Type    | `bool` |
| Default | `true` |

also install a timer that relaunches the cluster when the endpoint stops answering

### --watchdog-interval

|         |       |
| ------- | ----- |
| Type    | `int` |
| Default | `2`   |

minutes between health checks