Container isolation wasn't actually working. Sage was running two Hermes instances, cron jobs vanished into the wrong database, Telegram refused to send. I fixed all of that. Then the real bug turned out to be a Python loop that was never Docker's fault.

I ran three Hermes gateways as separate profiles on the same machine. Replies came from the wrong profile, Telegram connections fought each other. Here's why I containerised each one — and why that was only half the story.

After upgrading Unraid from 6.9 to 7.0.1, all my Docker containers became unmanageable - marked as '3rd party' and locked out of updates. Here's how I fixed it without recreating everything.