Fixed
Recreate crash on Gluetun-style stacks
Containers with network_mode: "container:gluetun" (or service:…) failed every update with:
docker: Error response from daemon: conflicting options: hostname and the network mode
…then rolled back. Cause: our recreate logic emitted --hostname and -p flags from the inspect data, but Docker forbids those when a container inherits another container's network namespace — they belong to the namespace owner. Now detects the container: / service: prefix and skips the conflicting flags.
Added
Container Groups: restart-dependents flag
Extension of the v1.15 Container Groups feature. A group can now be flagged "Restart dependents when the head container updates":
- The first container in the group is the "head" (e.g. Gluetun).
- When the head gets updated, all other members are restarted (not updated — just restarted) after the head reports healthy.
- Covers the VPN-sidecar workflow where every dependent shares the head's network namespace and loses connectivity when the namespace owner restarts.
- Includes a health-wait poll (up to the group's
wait_seconds) so dependents aren't kicked during the VPN handshake.
New checkbox on the group edit form (Advanced UI mode), persisted as restart_dependents: true in /data/groups.json.
Roadmap
Multi-host support is on the way as a v2.0 item — one Docksentry instance managing several Docker hosts, with per-host pending / history and hostname-prefixed notifications. Big enough to need its own release window. Wishlist input welcome via #2.
docker pull amayer1983/docksentry:latest
docker compose up -d