Niger Proxies

Keep each stage short and preserve request count, concurrency, connection reuse, response classes, bytes, and elapsed time. Stop at the first buyer-defined error or destination-control threshold; do not rotate around a refusal to continue. Pre-register the NER sample size, concurrency, timeout, retry allowance, and stop events. Then capture network, protocol, selector, exit, response category, bytes where relevant, and timestamp for each authorized attempt without dropping inconvenient rows. Use only an endpoint permitted for the Niger test, keep missing measurements explicit, and never rotate around a refusal or other mandatory stop.

Distinguish a gateway or connection failure from an HTTP redirect issued by the destination. A different final URL can also come from consent state, account data, language settings, or an experiment, so retain those inputs beside the chain. The adjacent Africa catalogue entries around Niger are Chad (15,827) and Liberia (15,092). These are historical totals under the same annual maintenance convention, so they are context for recorded breadth rather than substitutes for the Niger route test. Keep NER comparator rows labelled and diagnose the requested Niger route from its own evidence rather than a neighbouring count. Record why each comparator belongs in the authorized review before using it.

Compute the buyer-defined success rate from all planned attempts, with failed and inconclusive observations visible. Compare products only when their samples used the same labels, target, timeout, and stop rule. For the NER order, require a verified Niger exit, a maintained client and protocol, suitable session behavior, a declared valid-response rate, finite retries, and reviewable evidence. Recalculate after any configuration change instead of carrying forward an old pass. Reject any Niger conclusion built from an unverified exit, hidden failures, or a changed rule, and repeat NER verification after a material configuration change.