Drops per-run staging tables left behind by runs that have already finished.
Options
--yes- actually drop the tables--repo- the Ecto Repo to run against--prefix- the PostgreSQL schema GeoGenius is installed in
The pipeline drops a run's staging table in an after clause, and that drop
is deliberately rescued so a database failure during cleanup cannot destroy
the run's own failure record. A host whose database dies mid-cleanup
therefore leaks a table, nothing else reclaims it, and the schema down
migration's teardown refuses to drop a non-empty schema -- so a leak eventually blocks an
uninstall. This reclaims them.
A table whose import_run row is gone entirely is reclaimed too. That is
the case nothing else can reach: with no run row to read a status from,
such a table is invisible to anything that starts from import_run, and it
would otherwise sit in the schema until an operator found it by hand.
Like mix geo_genius.rollback, it prints what it would drop and does nothing
without --yes.