DC-NODE-43

DC derived declared

Gap-free resume of a reconnected live feed. A reconnected per-peer session resumes chain-sync from the last block actually DELIVERED downstream, never from the original boot anchor. Resuming from the boot anchor is WRONG, not merely slow: once the node has advanced more than k past it, the peer's rollback to that anchor exceeds the BLUE rollback k-guard and fails closed. Safety of the delivered-block resume point rests on lane ordering -- the per-peer lane is FIFO and is never dropped across a reconnect, so events buffered from the old session are delivered BEFORE the new session's; by the time the consumer processes the new session's rollback-to-intersection, that point is already durable and within k.

Source: docs/clusters/LIVE-WIRE-LIVENESS/SLICE-S1-keepalive-under-backpressure.md + docs/clusters/LIVE-WIRE-LIVENESS/SLICE-S2-live-feed-reconnect.md (INV-WL-7)

Enforcement trace

Code

crates/ade_node/src/node_lifecycle.rs

Tests 0

no tests named — gap

CI 0

no CI script — gap

Cross-references

Evidence notes

DECLARED, NOT ENFORCED: there is no mechanical test for the gap-free property -- proving it needs a synthetic N2N peer that drops and resumes while asserting no block is skipped, which would couple the test to the serve path. The argument above plus the 2026-08-01 live observation (resume from block 118907456 after a real peer restart, no k-guard violation, no rollback error) is EVIDENCE, not enforcement. Enforcement is owed. SUPPORTING live evidence only (never the reason for enforcement): a sustained preview run 2026-08-01 (~/.cardano-live1/ade-1376-s1s2.log) held the live tip for 8h+ with 1400 consecutive keep-alive round trips, max observed inter-line stall 20.0s (the cadence itself) against the peers ~97s limit, and recovered a deliberate docker restart of cardano-node-preview in 1m48s. Enforcement rests on the named tests + ci/ci_check_wire_liveness.sh, not on this run.