# #1436 v4tape emit lag (ppfeeds, 11 Oct)
Lag = emitted_t - block time, from swaps.jsonl (last 30k rows).
- Baseline p50 ~1.8 s, p90 ~2.5 s (calm hours). Spikes came from public/free RPC 429s: 10 Oct 17-18 UTC p90 2600-3000 s, 19-21 UTC p50 200-530 s, 22-23 UTC p90 60-150 s; calm again since 00 UTC. 1398 429 lines in v4tape.log.
- Cause: every call (getLogs, block ts, tx info) retried on 429 with backoff, and the tx-info batch (optional field) blocked the swap write.
- Fix (v4tape.py, backup .bak-1436, restarted 11 Oct 01:06 UTC, resumed from heartbeat, no gap): tx-info prefetch now tries=1, timeout 6 s; on failure tx_from/gas fields are None as before. Swap is no longer held by it.
- Not fixed: the 1.8 s baseline sits close to fast5's 2.5 s limit, so ~10% late entries remain. Next lever is a dedicated RPC for v4tape (own-node study #11 / paid research RPC needs exec/manager approval). fast5 could also measure lag from the swap's emitted_t only for the entry decision (testmgr call).