Worker Deployment and Diagnostics
Not yet drafted.
Chapter 8 covers the conceptual architecture of worker instances — one process, one database, one isolation surface, and the documented AllegroCache close/reopen incident worth avoiding. What's still needed is the operational half: the actual deployment scripts seen in the source zip but not yet walked through (run-all-ent.sh, restart-all-ent.sh, status-all-ent.sh, stop-ent.sh, the systemd install script ac-install-chandra-systemd.sh), what a healthy vs. unhealthy fleet looks like across chandra-worker-diagnostics.html's full surface (not just the identity-manifest piece already covered in Chapter 8), and how a new worker instance is actually provisioned end-to-end — from a Spine Editor packing decision (Chapter 13) to a running, registered (Chapter 7) instance.
This chapter is a strong candidate for the next drafting pass, since most of its raw material (the shell scripts, the full Worker Diagnostics tool) is already sitting in reviewed source and just hasn't been walked through operationally yet.