Late 2024 alternate install walkthrough

[farski]

In case this is helpful for anyone, I just rebuilt my birdnet box. I had been running birdnet-go for a while, but using the binaries, and on a Pi that had previously been running BirdNET-Pi. I started over, using Docker from the beginning. This is on a Pi 4B. I did things a bit differently than the standard install docs.

These instructions assume some basic Linux and Docker knowledge. I know not everyone installing birdnet-go does computers for a living, so if anything needs to be expanded, please ask – I'm happy to explain.

1. Installed Raspberry Pi OS Lite on an SD card using the RP imager
2. Installed Docker Engine by following their guide. (Note: a lot of links on Docker's website will take you to the install guide for Docker Desktop, even when you're on pages for Docker Engine. If you want only Docker Engine, make sure you know what page you're looking at).
3. Followed the steps to manage Docker as a non-root user
4. I made two directories for birdnet-go assets: ~/birdnet-go/data and ~/birdnet-go/config
5. Follow the steps in Deciding ALSA_CARD value
6. Created a ~/birdnet-go/compose.yml file

   services:
     birdnet-go:
       restart: unless-stopped
       image: ghcr.io/tphakala/birdnet-go:dev
       ports:
         - 80:8080
       environment:
         - ALSA_CARD=1
         - TZ=America/New_York
       devices:
         - /dev/snd:/dev/snd
       volumes:
         - $HOME/birdnet-go/config:/config
         - $HOME/birdnet-go/data:/data

   Note that this is running the dev release of birdnet-go, which may be more experimental or unstable than you want.

7. Run docker compose up -d. Because of the restart directive in compose.yml, Docker Compose should keep birdnet-go running, even in the event of a crash or something like that.
8. I also wanted to backup my birdnet database everyday, so I added this to my crontab: 15 0 * * * cp /home/birdnet/birdnet-go/data/birdnet.db /home/birdnet/birdnet-go/data/birdnet.bak.db && AWS_ACCESS_KEY_ID=tktktk AWS_SECRET_ACCESS_KEY="tktktk" AWS_DEFAULT_REGION=tktk aws s3 cp /home/birdnet/birdnet-go/data/birdnet.bak.db s3://my-backup-bucket/daily/birdnet.bak.db. You'd need the AWS CLI (I used sudo apt install awscli even though that's not what AWS recommends). You'll need to sort out creating a bucket in AWS and setting up the key/secret if you wanted to do this. Or you could do something similar with many other file hosting services, rsycn, sftp, etc. The only tricky bit in here is that aws s3 does an MD5 check to ensure the upload is valid, and the database file may change during the upload. So first this copies the DB to a backup file and pushes that to S3.

[ericleslie]

Thank you, this was very helpful!

[normal-dot-brad]

This is great and I'm close but I'm sorry I have a newbie question...
I've made the port: [mylocalip]:80:8080 but it's giving an error saying port 80 is already in use. Does it matter if I give it another number ?

[farski]

No, not really. Web browsers use port 80 by default, so if your birdnet box is running at 192.168.1.56, you would be able to access the UI by visiting just http://192.168.1.56. For any other port, the application will work just the same, but you'll have to include the port when you visit the site in your browser, like http://192.168.1.56:81 or http://192.168.1.56:8080.

I only use this Raspberry Pi for birdnet, so using port 80 makes it ever-so-slightly easier to get to.

[normal-dot-brad]

Ah ok, thank you that makes sense