Portugal Proxies

Disable automatic redirect following for the first pass, label each hop, and repeat with the same cookies and headers. Re-enable normal following only after the raw chain is stored so loops, cross-host moves, and transient failures remain visible. Number the PRT observations before starting. For each number, retain the authorized URL, route configuration, connection policy, exit check, response marker, elapsed time, and pass, fail, or inconclusive label. Do not manufacture extra attempts after the stop rule fires. Use only an endpoint permitted for the Portugal test, keep missing measurements explicit, and never rotate around a refusal or other mandatory stop.

Do not call every response difference geographic. A cookie, Accept-Language value, query parameter, edge node, deployment, or stale object may explain it; the log should show which key changed and whether the control changed too. In the maintained Europe list, Romania records 332,848 and Lithuania records 62,679, placing Portugal between them by historical count. That ordering says nothing about which route is currently online or suitable for the buyer's endpoint. Keep PRT comparator rows labelled and diagnose the requested Portugal route from its own evidence rather than a neighbouring count. Record why each comparator belongs in the authorized review before using it.

Choose the route for multi-step work only if the authorized flow completes within the stated window often enough for the buyer-defined threshold. Re-test after changing the client or session rule; do not silently change both. The buyer's PRT acceptance sheet should state required protocol, product, country observation rate, continuity or rotation rule, maximum retries, and valid-response evidence. The Portugal plan advances only when the pilot meets that sheet without excluding failures. Reject any Portugal conclusion built from an unverified exit, hidden failures, or a changed rule, and repeat PRT verification after a material configuration change.