Summary
Self-hosted open-data backfill can repeat the same GitHub GraphQL totals query across enqueue attempts and segment workers for a repository. During scheduled or manual backfill bursts, those repeated totals lookups add avoidable API pressure before each segment does its own work.
Desired outcome
Reuse fresh persisted totals snapshots for backfill planning, keep a valid same-source stale snapshot as a fallback when live refresh fails, and coalesce concurrent live totals refreshes for the same repository/source.
Impact
This should reduce duplicate GitHub GraphQL calls during backfill bursts, lower rate-limit risk for self-hosted installations, and keep segment expected-count decisions conservative when GitHub is unavailable.
Summary
Self-hosted open-data backfill can repeat the same GitHub GraphQL totals query across enqueue attempts and segment workers for a repository. During scheduled or manual backfill bursts, those repeated totals lookups add avoidable API pressure before each segment does its own work.
Desired outcome
Reuse fresh persisted totals snapshots for backfill planning, keep a valid same-source stale snapshot as a fallback when live refresh fails, and coalesce concurrent live totals refreshes for the same repository/source.
Impact
This should reduce duplicate GitHub GraphQL calls during backfill bursts, lower rate-limit risk for self-hosted installations, and keep segment expected-count decisions conservative when GitHub is unavailable.