# ratelimit #2950 (ppfeeds, 18 Oct)
- Lines: about 11 per hour, all "feed: both RPCs failed: HTTP 429" in wlink, wlink1 and mc5 (loop_err / late_scan_err). Read from /opt/fm_mirror.
- Cause: the dRPC key is refused (403, Ben decision #2710), so the firstmin bots run on the public node only. When it returns 429, the second RPC is the same public limit, so "both RPCs failed". wlink, wlink1, mc5 and l2v share one public IP with no shared throttle or cache.
- Fix at the cause: (1) Ben restores the dRPC plan (#2710). (2) exec: add the picker-box relay (other IP) as a third endpoint and jittered 429 backoff; ideally one shared RPC cache for the three bots.
- Not changed: firstmin (out of reach for me). Same finding earlier: ppfeeds/ratelimit-2546.md.