Soft and hard restart

Version: 0.2 · Last Updated: 2026-09-08 · Status: 🔴 DA REVISIONARE

The need. The server must be restartable — urgently or gently — without betraying the people working on it at that moment.

Stopping and restarting a living server without losing what must survive: the restart liturgy (hard/soft at the decider's choice, notice via notify_user, delegation to ServerApplication, execv) and dump/restore across a full server restart.

Interactions: orchestration (park everybody, refill) · sessions (their snapshot exists already) · global-store (explicitly NOT restored — owner decision 2026-08-22).

Server restart and orchestration recovery

Server restart owns stop/serve, execv and soft boot of the base. Orchestration and deployment add their own recovery mechanisms: the SPA world adds parking the users (freeze, refill), subcommanders add branch reconstruction, Kubernetes adds the Pod lifecycle.