Drop dead ores_controller_service_instances_tbl/events_tbl
Table of Contents
This page is a capture in the inbox bucket of the product backlog — a pre-sprint idea, not yet pulled into a sprint as a story.
What
Drop ores_controller_service_instances_tbl and
ores_controller_service_events_tbl
(projects/ores.sql/create/controller/,
projects/ores.sql/drop/controller/) along with their create/drop
SQL and any populate scripts. A full-repo grep for both table names
(excluding build/ and venv/) found zero readers or writers left
in C++, Python, or SQL – they were the controller's own Kubernetes-
style Pod/Event-log bookkeeping (PID, phase, restart count, event
history) for processes it launched and supervised itself; systemd now
owns that role via its own unit state and journal, so nothing
populates or queries these tables anymore.
Why
ores.controller.service=/=process_supervisor was fully decommissioned
in Sprint 24 (PR #1806), but its DB schema was left behind. Unlike
ores_controller_service_definitions_tbl=/=dependencies_tbl –
which stayed genuinely load-bearing as the data source
systemd_generate.py's fetch_service_definitions()=/
=fetch_dependencies() read from – service_instances_tbl and
service_events_tbl have no equivalent successor consumer. Leaving
dead tables (one a TimescaleDB hypertable with its own retention
policy) in the schema is pure confusion for anyone reading it cold,
and quiet drift risk if something is ever mistakenly pointed at them
again.
References
- projects/ores.sql/create/controller/controller_service_instances_create.sql
- projects/ores.sql/create/controller/controller_service_events_create.sql
- projects/ores.sql/drop/controller/controller_service_instances_drop.sql
- projects/ores.sql/drop/controller/controller_service_events_drop.sql
- projects/ores.compass/src/systemd_generate.py (the two tables that are not dead – definitions/dependencies – for contrast)
See also
- Drop controller_service_* tables, move service registry to codegen model – the sibling capture for the two definitions/dependencies tables that are still in active use; that one needs a codegen migration first, this one can just be dropped.
- Task: Investigate decommissioning ores.controller.service and its DB tables – the audit task that found this.