Improve your public instance performance and stability

This tutorial has been written by unixfox, owner of yewtu.be. He is better suited when looking for help about this tutorial.

Synopsis

This tutorial will explain how to improve the performance, and decrease the memory and CPU consumption of your public instance on your server.

Using Docker is recommended for this tutorial because the process is simpler with it. You can try it outside of Docker by adapting the ideas of this tutorial, but the main focus will be Docker.

Instructions

1) Tune up your config.yml

For decreasing the load on the PostgreSQL database:

enable_user_notifications: false
User notifications will be deactivated.
On large instances, it is recommended to set this option to false in order to reduce the amount of data written to the database, and hence improve the overall performance of the instance.

2) Multiple Invidious processes

Invidious is single threaded, so by running multiple processes you better utilize the multiple threads of your server.
Also past a certain amount of requests, Invidious becomes sluggish. Having multiple processes reduces this sluggishness.

This tutorial will explain how to run multiple containers of Invidious and use NGINX in Docker as a loadbalancer.
If your reverse proxy is already in Docker you can also adapt it so that it sends the requests directly to the Invidious containers.

We assume that you have not changed the port 3000 from the default installation. But if you changed it then you will need to adapt it in the following configuration files and commands.

In your docker-compose.yml, remove these lines for binding the port of Invidious:
```
ports:
 - "127.0.0.1:3000:3000"
```
Duplicate the whole "invidious service" and copy it into a new one named invidious-refresh.
Yes the whole block from the "invidious service". From invidious: to - invidious-db.
Add these two lines into your INVIDIOUS_CONFIG parameter for only the invidious service:
```
channel_threads: 0
feed_threads: 0
```
This is required so that only one invidious process refresh the subscriptions for the users.
Running this process with multiple processes may introduce some conflicts.
Edit the docker-compose.yml with this content:
```
version: "3"
services:
   invidious:
      image: quay.io/invidious/invidious:latest
      deploy:
         replicas: 6
```
Explanation: The deploy.replicas parameter allows running multiple containers of the same Docker image.
Note: You can set more or less Invidious processes (6 in the example).
Don't restart Invidious yet!

Create a file called nginx.conf and add this content:

user www-data;
events {
     worker_connections 1000;
}
http {
     server {
         listen 3000;
         listen [::]:3000;
         access_log off;
         location / {
             resolver 127.0.0.11;
             set $backend "invidious";
             proxy_pass http://$backend:3000;
             proxy_http_version 1.1; # to keep alive
             proxy_set_header Connection ""; # to keep alive
         }
     }
}

Add a new service in your docker-compose.yml file:

nginx:
   image: nginx:latest
   restart: unless-stopped
   volumes:
     - ./nginx.conf:/etc/nginx/nginx.conf:ro
   depends_on:
     - invidious
   ports:
     - "127.0.0.1:3000:3000"

Update your cronjobs to restart Invidious (if you use cron). Instead of restarting a single Docker container you will now need to restart 6 containers (adjust if you added or removed number of containers). Replace with these CRON lines:
```
0 */1 * * * docker restart invidious-invidious-1
1 */1 * * * docker restart invidious-invidious-2
2 */1 * * * docker restart invidious-invidious-3
3 */1 * * * docker restart invidious-invidious-4
4 */1 * * * docker restart invidious-invidious-5
5 */1 * * * docker restart invidious-invidious-6
```
Each CRON line has a different schedule to avoid disrupting your entire Invidious instance due a restart.
Apply the new configuration:
```
docker compose down
docker compose up
```

??? note "Click here for a final example of the docker-compose file. (Don't copy blindly)"

```yaml
version: "3"
services:
    invidious:
        image: quay.io/invidious/invidious:latest
        deploy:
            replicas: 6
        restart: unless-stopped
        environment:
            INVIDIOUS_CONFIG: |
                channel_threads: 0
                feed_threads: 0
                db:
                    dbname: invidious
                    user: kemal
                    password: kemal
                    host: invidious-db
                    port: 5432
                check_tables: true
                use_pubsub_feeds: true
                use_innertube_for_captions: true
                external_port: 443
                domain: mydomain.com
                https_only: true
                statistics_enabled: true
                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

    
    invidious-refresh:
        image: quay.io/invidious/invidious:latest
        restart: unless-stopped
        environment:
            INVIDIOUS_CONFIG: |
                db:
                    dbname: invidious
                    user: kemal
                    password: kemal
                    host: invidious-db
                    port: 5432
                check_tables: true
                use_pubsub_feeds: true
                use_innertube_for_captions: true
                external_port: 443
                domain: mydomain.com
                https_only: true
                statistics_enabled: true
                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 

    nginx:
        image: nginx:latest
        restart: unless-stopped
        volumes:
            - ./nginx.conf:/etc/nginx/nginx.conf:ro
        depends_on:
            - invidious
        ports:
            - "127.0.0.1:3000:3000"

    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:
```

3) Speed up video playback with http3-ytproxy

Kavin from the Piped team has developed a tool that is much faster at proxying the traffic for video playback and image loading than Invidious: https://github.com/TeamPiped/http3-ytproxy

NGINX configuration will be used for this tutorial, and it's highly recommended to setup this configuration on your main reverse proxy.

But if you do not have NGINX as your main reverse proxy you can either try to adapt the rules to your reverse proxy. Or you can also use the separate NGINX container that you created in the second section.

Find the username of NGINX process, you can find it at the top of the file /etc/nginx/nginx.conf.
Get its uid and gid using the id command like so: id www-data.
Usually it's uid 33 and gid 33.
Create a new directory for this new service and give the correct permissions for the user of NGINX:
```
mkdir /opt/http3-ytproxy
chown 33:33 -R /opt/http3-ytproxy
```
Replace 33:33 with uid:gid if you have something different.

Add this new service to your docker-compose.yml:

http3-ytproxy:
   image: 1337kavin/ytproxy:latest
   restart: unless-stopped
   user: 33:33
   network_mode: "host"
   environment:
     - DISABLE_WEBP=1
   volumes:
     - /opt/http3-ytproxy:/app/socket

Replace 33:33 with uid:gid if you have something different.

Add these lines for the "server" section of Invidious in your NGINX configuration, just after the block location /:

location ~ (^/videoplayback|^/vi/|^/ggpht/) {
     proxy_buffering on;
     proxy_buffers 1024 16k;
     proxy_set_header X-Forwarded-For "";
     proxy_set_header CF-Connecting-IP "";
     proxy_hide_header "alt-svc";
     sendfile on;
     sendfile_max_chunk 512k;
     tcp_nopush on;
     aio threads=default;
     aio_write on;
     directio 16m;
     proxy_hide_header Cache-Control;
     proxy_hide_header etag;
     proxy_http_version 1.1;
     proxy_set_header Connection keep-alive;
     proxy_max_temp_file_size 32m;
     access_log off;
     proxy_pass http://unix:/opt/http3-ytproxy/http-proxy.sock;
     add_header Cache-Control private always;
}

Note: proxy_buffering on may increase the disk usage, if this is not what you want you can remove the two lines proxy_buffering and proxy_buffers.
If you're using the NGINX from the second section, you will need to add this new volume:

volumes:
 - /opt/http3-ytproxy:/opt/http3-ytproxy

Reload the docker compose. (if you followed the second section)
Reload NGINX: systemctl reload nginx.