[BUG] docker-compose: otel-collector/prometheus/grafana take dynamic IPs from 10.10.10.0/24 and can grab 10.10.10.2, nostream fails with "Address already in use" #929

Describe the bug
docker-compose.yml pins nostream, nostream-db and nostream-cache to 10.10.10.2–10.10.10.4, but otel-collector, prometheus and grafana (added with the observability stack) take dynamic addresses from the same 10.10.10.0/24 pool. Docker hands out dynamic addresses from the bottom of the subnet, so whichever of them is created first grabs 10.10.10.2 and nostream then fails:

Error response from daemon: failed to set up container networking: Address already in use

otel-collector has no dependencies, so it is usually created before nostream; it also happens after a host reboot, when the daemon starts containers in arbitrary order.

To Reproduce

1. docker compose down (network removed), then docker compose up -d.
2. docker network inspect nostream → otel-collector=10.10.10.2/24; nostream stays in Created with the error above.

Expected behavior
Dynamic addresses never collide with the static ones.

System

• OS: Ubuntu 22.04, Docker 29.9, compose v5.4
• Version: 3.4.1

Fix (running in production)
Restrict dynamic allocation to a range outside the static addresses; subnet and gateway stay the same (10.10.10.1 is in the default trustedProxies):

networks:
  default:
    name: nostream
    ipam:
      driver: default
      config:
        - subnet: 10.10.10.0/24
          gateway: 10.10.10.1
          ip_range: 10.10.10.128/26

This also keeps .5 (docker-compose.local.yml) and .252/.253 (i2p/tor overlays) out of the dynamic pool. After this change: nostream=10.10.10.2, otel-collector=10.10.10.128, prometheus=10.10.10.129, grafana=10.10.10.130. Happy to send it as a PR, but PR creation on this repo is limited for me (see #927).