{"id":"wal-durability-tiers-lost-during-recovery","text":"The WAL's carefully calibrated write-time durability tiers (per-write fsync in sync mode vs. batch-only fsync) are completely invisible to recovery: sequence numbers and checkpoints that could distinguish between definitely-durable and potentially-lost records are never consulted, so replay treats all CRC-valid records identically regardless of whether they were fsynced — the write-time investment in durability is wasted.","truth_value":"OUT","source":"","source_url":"","source_hash":"","justifications":[{"type":"SL","antecedents":["wal-two-tier-durability-model","wal-recovery-infrastructure-is-vestigial"],"outlist":[],"label":"Write-time durability calibration is invisible to the recovery path that ignores its own infrastructure"}],"dependents":["write-time-durability-engineering-abandoned-at-recovery"],"metadata":{},"created_at":"","updated_at":"","reviewed_at":"","verified_at":"","retracted_at":"","explanation":{"steps":[{"node":"wal-durability-tiers-lost-during-recovery","truth_value":"OUT","reason":"SL justification invalid","failed_antecedents":["wal-recovery-infrastructure-is-vestigial"],"label":"Write-time durability calibration is invisible to the recovery path that ignores its own infrastructure"}]}}