# ratelimit #2546 (ppfeeds, 16 Oct)
- firstmin 429 lines (mc5 25, wl_wlink 24, wl_wlink1 20) are ONE burst: 00:26-00:30 BKK (17:26-17:30 UTC), plus a few at 00:47-00:54. Since then about 1 line in 1.5 h (mirror logs).
- Same minutes, the public node returned 429 on every box (aipick, laneradar, v4tape, pretag) and GeckoTerminal failed on a 3-min schedule. The provider throttled the whole public endpoint; it is not a duplicate caller on firstmin.
- Our side was already lowered: rpcpool SHARED_GAP 0.12 and solidrpc batch skip (#2373/#2383). The burst did not repeat.
- Not changed: firstmin code (out of reach). Suggestion for exec: firstmin bots use only 2 RPCs ("both RPCs failed"); a third fallback endpoint or jittered backoff on 429 would cover such bursts. No money loss seen (loop_err retries).