# #3229 RPC (public)@firstmin-trader 429s (18 Oct 23:45 BKK)
Evidence (/opt/fm_mirror logs): "loop_err RuntimeError(feed: both RPCs failed: HTTPError 429)" in mc5 + wl_wlink, wl_wlink1, wl_cleane, wl_v3early.
Pattern: bursts, e.g. 23:18-23:19 = 69 lines in 2 min across all bots at once, then ~1/min. Same-minute hits on every bot = all bots poll the same two public RPCs on the same cadence from one IP, with no shared throttle/cache between processes. Both RPCs 429 together, so fallback/retry does not help.
Cause is inside firstmin (feed code); ppfeeds cannot reach or edit it. Not research-box load (different IP).
Suggested fix for exec, at the cause: one shared per-host pacer (file lock / token bucket, like rpcfall._pace gap 0.12s on research-box) in the feed's rpc call, plus a shared head-block cache so the bots make one call per block instead of one each. No retry loops.
Impact seen: loop_err skips a cycle only; no failed trades in these logs.