documentation/docs/installation.md

# Installation

After installation take a look at the [Post-install steps](#post-install-configuration).

Note: Any [PaaS](https://en.wikipedia.org/wiki/Platform_as_a_service) or [SaaS](https://en.wikipedia.org/wiki/Software_as_a_service) provider/software (Heroku, YunoHost, Repli...) are unsupported. Use them at your own risk. They **WILL** cause problems with Invidious and might even suspend your account for "abuse" since Invidious is heavy, bandwidth intensive and technically a proxy (and most providers don't like them). If you use one and want to report an issue, please mention which one you use.


## Hardware requirements

Running Invidious requires at least 20GB disk space, 512MB of free RAM (so ~2G installed on the system), as long as it is restarted regularly, as recommended in the post-install configuration. Public instances should ideally have at least 60G disk space, 4GB of RAM, 2vCPU, a 200 mbps link and 20TB of traffic (no data cap/unlimited traffic is preferred).

Compiling Invidious requires at least 2.5GB of free RAM (We recommend to have at least 4GB installed).
If you have less (e.g on a cheap VPS) you can setup a SWAP file or partition, so the combined amount is >= 4GB.

You need at least 1GB of RAM for the machine that will run the tool `youtube-trusted-session-generator` in the 1st step. Doesn't need to be the same machine as the one running Invidious, just a machine running on the same public IP address.

## Docker

**The Invidious docker image is only [available on Quay](https://quay.io/repository/invidious/invidious) because, unlike Docker Hub, [Quay is Free and Open Source Software](https://github.com/quay/quay/blob/master/LICENSE). This is reflected in the `docker-compose.yml` file used in this walk-through.**

Ensure [Docker Engine](https://docs.docker.com/engine/install) and [Docker Compose](https://docs.docker.com/compose/install) are installed before beginning.

### Docker-compose method (production)

**This method uses the pre-built Docker image from quay**

Note: Currently the repository has to be cloned, this is because the `init-invidious-db.sh` file and the `config/sql` directory have to be mounted to the postgres container (See the volumes section in the docker-compose file below). This "problem" will be solved in the future.

Make sure to run the newer Docker Compose V2: https://docs.docker.com/compose/install/linux/. It should already be installed if you can successfully run the command `docker compose` (with a space between the two words).

??? warning "About po_token and visitor_data identities"

    po_token known as Proof of Origin Token. This is an attestation token generated by a complex anti robot verification system created by Google named BotGuard/DroidGuard. It is used to confirm that the request is coming from a genuine device.
    
    These identity tokens (po_token and visitor_data) generated in this tutorial will make your entire Invidious session more easily traceable by YouTube because it is tied to a unique identifier.
    
    There is currently no official automatic tool to periodically change these tokens. This is working in progress but, for the time being, this is the solution the Invidious team is offering.

    If you want to be less traceable, you can always script the process by changing the identities every X hour.


1.  Generate po_token and visitor_data identities for passing all verification checks on YouTube side:
    ```
    docker run quay.io/invidious/youtube-trusted-session-generator
    ```
    You have to run this command on the same public IP address as the one blocked by YouTube. Not necessarily the same machine, just the same public IP address.  
    You will need to copy these two parameters in the third step.  
    Subsequent usage of this same token will work on the same IP range or even the same ASN. The point is to generate this token on a blocked IP as "unblocked" IP addresses seems to not generate a token valid for passing the checks on a blocked IP.  

2.  Execute these commands:
    ```bash
    git clone https://github.com/iv-org/invidious.git
    cd invidious
    ```

3.  Edit the docker-compose.yml with this content:

    ```docker
    version: "3"
    services:

      invidious:
        image: quay.io/invidious/invidious:latest
        # image: quay.io/invidious/invidious:latest-arm64 # ARM64/AArch64 devices
        restart: unless-stopped
        ports:
          - "127.0.0.1:3000:3000"
        environment:
          # Please read the following file for a comprehensive list of all available
          # configuration options and their associated syntax:
          # https://github.com/iv-org/invidious/blob/master/config/config.example.yml
          INVIDIOUS_CONFIG: |
            db:
              dbname: invidious
              user: kemal
              password: kemal
              host: invidious-db
              port: 5432
            check_tables: true
            signature_server: inv_sig_helper:12999
            visitor_data: CHANGE_ME
            po_token: CHANGE_ME
            # external_port:
            # domain:
            # https_only: false
            # statistics_enabled: false
            hmac_key: "CHANGE_ME!!"
        healthcheck:
          test: wget -nv --tries=1 --spider http://127.0.0.1:3000/api/v1/trending || exit 1
          interval: 30s
          timeout: 5s
          retries: 2
        logging:
          options:
            max-size: "1G"
            max-file: "4"
        depends_on:
          - invidious-db

      inv_sig_helper:
        image: quay.io/invidious/inv-sig-helper:latest
        init: true
        command: ["--tcp", "0.0.0.0:12999"]
        environment:
          - RUST_LOG=info
        restart: unless-stopped
        cap_drop:
          - ALL
        read_only: true
        security_opt:
          - no-new-privileges:true

      invidious-db:
        image: docker.io/library/postgres:14
        restart: unless-stopped
        volumes:
          - postgresdata:/var/lib/postgresql/data
          - ./config/sql:/config/sql
          - ./docker/init-invidious-db.sh:/docker-entrypoint-initdb.d/init-invidious-db.sh
        environment:
          POSTGRES_DB: invidious
          POSTGRES_USER: kemal
          POSTGRES_PASSWORD: kemal
        healthcheck:
          test: ["CMD-SHELL", "pg_isready -U $$POSTGRES_USER -d $$POSTGRES_DB"]

    volumes:
      postgresdata:
    ```

    Note: This compose is made for a true "production" setup, where Invidious is behind a reverse proxy. If you prefer to directly access Invidious, replace `127.0.0.1:3000:3000` with `3000:3000` under the `ports:` section.

4. Run the docker composition:

```
docker compose up -d
```


### Docker-compose method (development)

**This method builds a Docker image from source**

```bash
git clone https://github.com/iv-org/invidious.git
cd invidious
docker-compose up
```


## Manual Installation

### Linux

#### Generate po_token and visitor_data identities

[Follow these instructions here on the official tool `youtube-trusted-session-generator`](https://github.com/iv-org/youtube-trusted-session-generator?tab=readme-ov-file#tutorial-without-docker)

These two parameters will be required for passing all verification checks on YouTube side and you will have to configure them in Invidious.

You have to run this command on the same public IP address as the one blocked by YouTube. Not necessarily the same machine, just the same public IP address.  
You will need to copy these two parameters in the `config.yaml` file.  
Subsequent usage of this same token will work on the same IP range or even the same ASN. The point is to generate this token on a blocked IP as "unblocked" IP addresses seems to not generate a token valid for passing the checks on a blocked IP.  

??? warning "About po_token and visitor_data identities"

    po_token known as Proof of Origin Token. This is an attestation token generated by a complex anti robot verification system created by Google named BotGuard/DroidGuard. It is used to confirm that the request is coming from a genuine device.
    
    These identity tokens (po_token and visitor_data) generated in this tutorial will make your entire Invidious session more easily traceable by YouTube because it is tied to a unique identifier.
    
    There is currently no official automatic tool to periodically change these tokens. This is working in progress but, for the time being, this is the solution the Invidious team is offering.

    If you want to be less traceable, you can always script the process by changing the identities every X hour.


#### Run inv_sig_helper in background

[Follow these instructions here on the official tool `inv_sig_helper`](https://github.com/iv-org/inv_sig_helper?tab=readme-ov-file#building-and-running-without-docker) and run it in the background with systemd for example.

inv_sig_helper handle the "deciphering" of the video stream fetched from YouTube servers. As it is running untrusted code from Google themselves, make sure to isolate it by for example running it inside a LXC or locked down through systemd.

The following systemd service file can be used to run inv_sig_helper with systemd: https://github.com/iv-org/inv_sig_helper/blob/master/inv_sig_helper.service

#### Install Crystal

Follow the instructions for your distribution here: https://crystal-lang.org/install/

**Note:** Invidious currently supports the following Crystal versions: `1.10.x` / `1.11.x` / `1.12.x`. \
Versions `1.9.x` and older are incompatible because we use features only present in the newer versions. \
Versions `1.13.x` should be compatible, however we did not test it.

#### Install the dependencies

Arch Linux
```bash
sudo pacman -S base-devel librsvg postgresql ttf-opensans
```

Debian/Ubuntu
```bash
sudo apt install libssl-dev libxml2-dev libyaml-dev libgmp-dev libreadline-dev postgresql librsvg2-bin libsqlite3-dev zlib1g-dev libpcre3-dev libevent-dev fonts-open-sans
```

RHEL based and RHEL-like systems (RHEL, Fedora, AlmaLinux, RockyLinux...)
```bash
sudo dnf install -y openssl-devel libevent-devel libxml2-devel libyaml-devel gmp-devel readline-devel postgresql librsvg2-devel sqlite-devel zlib-devel gcc  open-sans-fonts
```

#### Add an Invidious user and clone the repository

```bash
useradd -m invidious
su - invidious
git clone https://github.com/iv-org/invidious
exit
```

#### Set up PostgreSQL

```bash
systemctl enable --now postgresql
sudo -i -u postgres
psql -c "CREATE USER kemal WITH PASSWORD 'kemal';" # Change 'kemal' here to a stronger password, and update `password` in config/config.yml
createdb -O kemal invidious
exit
```

#### Set up Invidious

```bash
su - invidious
cd invidious
make

# Configure config/config.yml as you like
cp config/config.example.yml config/config.yml 

# edit config.yaml to include po_token and visitor_data previously generated

edit config/config.yaml

# Deploy the database
./invidious --migrate

exit
```

#### Systemd service

```bash
cp /home/invidious/invidious/invidious.service /etc/systemd/system/invidious.service
systemctl enable --now invidious.service
```

### MacOS

#### Generate po_token and visitor_data identities

[Follow these instructions here on the official tool `youtube-trusted-session-generator`](https://github.com/iv-org/youtube-trusted-session-generator?tab=readme-ov-file#tutorial-without-docker)

These two parameters will be required for passing all verification checks on YouTube side and you will have to configure them in Invidious.

You have to run this command on the same public IP address as the one blocked by YouTube. Not necessarily the same machine, just the same public IP address.  
You will need to copy these two parameters in the `config.yaml` file.  
Subsequent usage of this same token will work on the same IP range or even the same ASN. The point is to generate this token on a blocked IP as "unblocked" IP addresses seems to not generate a token valid for passing the checks on a blocked IP.  

??? warning "About po_token and visitor_data identities"

    po_token known as Proof of Origin Token. This is an attestation token generated by a complex anti robot verification system created by Google named BotGuard/DroidGuard. It is used to confirm that the request is coming from a genuine device.
    
    These identity tokens (po_token and visitor_data) generated in this tutorial will make your entire Invidious session more easily traceable by YouTube because it is tied to a unique identifier.
    
    There is currently no official automatic tool to periodically change these tokens. This is working in progress but, for the time being, this is the solution the Invidious team is offering.

    If you want to be less traceable, you can always script the process by changing the identities every X hour.

#### Run inv_sig_helper in background

[Follow these instructions here on the official tool `inv_sig_helper`](https://github.com/iv-org/inv_sig_helper?tab=readme-ov-file#building-and-running-without-docker)

inv_sig_helper handle the "deciphering" of the video stream fetched from YouTube servers. As it is running untrusted code from Google themselves, make sure to isolate it by for example running it inside Docker or a VM.

Call for action: An example here is welcome, [if you want to contribute to one](https://github.com/iv-org/documentation/edit/master/docs/installation.md#macos).

#### Install the dependencies

```bash
brew update
brew install crystal postgresql imagemagick librsvg
```

#### Clone the Invidious repository

```bash
git clone https://github.com/iv-org/invidious
cd invidious
```

#### Set up PostgreSQL

```bash
brew services start postgresql
createdb
psql -c "CREATE ROLE kemal WITH LOGIN PASSWORD 'kemal';" # Change 'kemal' here to a stronger password, and update `password` in config/config.yml
createdb -O kemal invidious
psql invidious kemal < config/sql/channels.sql
psql invidious kemal < config/sql/videos.sql
psql invidious kemal < config/sql/channel_videos.sql
psql invidious kemal < config/sql/users.sql
psql invidious kemal < config/sql/session_ids.sql
psql invidious kemal < config/sql/nonces.sql
psql invidious kemal < config/sql/annotations.sql
psql invidious kemal < config/sql/playlists.sql
psql invidious kemal < config/sql/playlist_videos.sql
```

#### Set up Invidious

```bash
make

# Configure config/config.yml as you like
cp config/config.example.yml config/config.yml

# edit config.yaml to include po_token and visitor_data previously generated

edit config/config.yaml
```

### Windows

Crystal, the programming language used by Invidious, [doesn't officially support Windows yet](https://github.com/crystal-lang/crystal/issues/5430) but you can still install Invidious:

- By installing [Docker desktop](https://docs.docker.com/desktop/install/windows-install/) and then following [our guide about Docker](#docker).
- By installing [Windows Subsystem for Linux](https://msdn.microsoft.com/en-us/commandline/wsl/about) and then following [our guide about Linux](#linux).
- By installing [Windows-specific builds](https://github.com/crystal-lang/crystal/releases/) of Crystal. Be wary, as we don't currently have records of Invidious being tested on those "unsupported" builds yet.

## Post-install configuration:

Detailed configuration available in the [configuration guide](./configuration.md).

You must set a random generated value for the parameter `hmac_key:`! On Linux you can generate it using the command `pwgen 20 1`.

Because of various issues, Invidious **must** be restarted often, at least once a day, ideally every hour.

If you use a reverse proxy, you **must** configure invidious to properly serve request through it:

`https_only: true` : if you are serving your instance via https, set it to true

`domain: domain.ext`: if you are serving your instance via a domain name, set it here

`external_port: 443`: if you are serving your instance via https, set it to 443

`use_pubsub_feeds: true`: if you are serving your instance on the internet, allow for faster notification of new videos ([detailed explanation](https://github.com/iv-org/invidious/blob/97c4165f55c4574efb554c9dae8d919d08da1cdd/config/config.example.yml#L409)).

`use_innertube_for_captions: true`: if you are serving a public instance or you are hosting invidious in a datacenter, allow to unblock captions ([detailed explanation](https://github.com/iv-org/invidious/issues/2567#issuecomment-1727928996)).

## Update Invidious

#### Updating a Docker install
```bash
docker-compose pull
docker-compose up -d
docker image prune -f
```

#### Update a manual install
```bash
su - invidious
cd invidious
git pull
make
exit
systemctl restart invidious.service
```

## Usage:

```bash
./invidious
```


#### Logrotate configuration

```bash
echo "/home/invidious/invidious/invidious.log {
rotate 4
weekly
notifempty
missingok
compress
minsize 1048576
}" | tee /etc/logrotate.d/invidious.logrotate
chmod 0644 /etc/logrotate.d/invidious.logrotate
```