pub fn restore_in_progress(db: &Db) -> Result<bool>Expand description
Whether any embedded-cohort pipeline has a bulk restore in
progress: an __restore row in the InProgress state, meaning a
restore crashed mid-run and its per-shard cursors are resumable.
Tip indexing never writes __restore rows, and the restore
driver’s finalize flips every registered row to Complete in one
atomic batch, so an InProgress row is unambiguous. The embedded
bootstrap uses this to tell a resumable restore apart from other
states that merely lack live watermarks – a fresh store, or a
from-genesis run that crashed between the live pipelines’ first
commits – which must be cleared before restoring, because the
bulk restore merges and puts on top of whatever rows exist.