NAS / DOCKER / IPV6 / RECOVERY
NAS Services and Docker Networking
The NAS does more than store files. It runs some of the services my network depends on, so a missing IPv6 address or a wrong container route can cause problems well beyond the NAS itself.
What runs there
The NAS hosts Docker-based infrastructure such as AdGuard, a reverse proxy, the UniFi controller, remote-support services, password management, PXE/ISO boot support and monitoring. I also run other applications on it, but this project is about their network connections and how they come back after maintenance.
Two networking problems I had to fix
Missing IPv6 ULA. Clients were relying on a stable internal IPv6 address for DNS, but that address could disappear after interface changes. My original hook was not reliable enough. I replaced it with an idempotent recovery script, a systemd oneshot service and a timer, then checked the address after the relevant events.
Docker routing. A reverse-proxy container attached to multiple networks could end up with an unsuitable default route or IPv6 address. I corrected the network and gateway preferences in the Compose configuration so that recreating the container wouldn't put the old problem back.
Updates without losing the infrastructure
When updating the NAS operating system and Docker, I need more than a list of container names. I keep copies of Compose definitions, environment files, volumes, network settings, restart policies and supporting scripts. DNS is brought back early, followed by the reverse proxy and dependent services.
I tested this maintenance sequence during a real upgrade. It helped me spot service-order dependencies that are easy to miss if everything is simply set to restart automatically.
What I want to check next
The services have been brought back after maintenance, but I still want to restore selected applications from independent backups rather than only verify that their containers start. I also want to keep testing the NAS's IPv6 recovery after network changes.