Keeping self hosted services updated...

... for the reckless. I have a few services running locally. Keeping them updated is getting more and more bothersome. There are a few things to keep in mind when using Docker images: Do not use latest - you want to pin the version you are deploying and read the changelog / update notes before upgrading. More importantly, do not update without having read the changelog / update notes.

But when your ~/compose directory looks like this

timo@alextrasza:~/compose$ ls
acme     drawio      grafana        jellyfin    memos      ollama     prometheus  searxng   tavern         wyoming-whisper
ampache  excalidraw  homarr         komga       nextcloud  openwebui  roundcube   sshwifty  vaultwarden    yarr
caddy    floatplane  homeassistant  linkwarden  nodered    photoview  roundcube2  stalwart  wyoming-piper

this process becomes a good amount of work. This is one of the instances where I implore you to do as I say, not as I do. I've been self hosting things for a long time. I have nightly backups and usually kick off a backup before updating, so I can roll back easily if I have to. I also only run updates when I can spare 30 minutes to do a rollback if needed.

Let me tell you a secret - it was not necessary in the last 48 months to ever do a restore from a backup. Projects got really good at avoiding breaking changes and at having a clean migration path when running latest. This goes very specifically for the projects I host myself - your mileage may vary. So, one more time: do not do the following in production with services you need to be online and available 24/7:

#!/bin/bash

BASE_PATH="/home/timo/compose"
ORIGINAL_DIR=$(pwd)

cd "$BASE_PATH" || { echo "Failed to enter directory: $BASE_PATH"; exit 1; }

for dir in */; do
  if [ -d "$dir" ]; then
    cd "$dir" || { echo "Failed to enter subdirectory: $dir"; continue; }

    echo "Updating $dir"

    docker compose pull
    docker compose up -d

    cd - > /dev/null || { echo "Failed to return to parent directory"; continue; }
  fi
done

cd "$ORIGINAL_DIR" || { echo "Failed to return to original directory"; exit 1; }

echo "All compose setups updated."

This little script saves me a lot of typing, changing directories and running the same two commands repeatedly. But it also only works on my home network service host. Production follows best practices with tagged images, proper changelog review and updates outside of business hours.