Romania Proxies

Hold the endpoint, protocol, credentials, and request constant while reopening the transport according to the documented client behavior. Report unique exits, repeats, failed observations, and sample size together; requests multiplexed on one connection are not independent rotations. Make the ROU log append-only during the pilot. Capture permission basis, intended selector, network and protocol, connection boundary, exit result, target response class, bytes where relevant, and UTC time. Apply the original stop condition without rewriting earlier rows. Use only an endpoint permitted for the Romania test, keep missing measurements explicit, and never rotate around a refusal or other mandatory stop.

A local refusal, proxy authentication error, SOCKS handshake problem, TLS error, and destination denial belong to different owners. Preserve the earliest failing boundary and correct that layer before repeating the country test. Romania sits beside Poland (373,924) and Portugal (101,316) in the Europe historical-count ordering. Preserve the country labels. Combining their observations would erase the distinction the buyer is trying to validate. Keep ROU comparator rows labelled and diagnose the requested Romania 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 browser QA only when a second clean run reproduces the accepted observation and the exit country is independently verified. Retain the profile recipe so another reviewer can repeat it. The Romania decision is ready only when ROU evidence covers country verification, application compatibility, session behavior, response classification, usage, and stop conditions. A pilot that changes endpoints or acceptance rules midway must be rerun or labelled inconclusive. Reject any Romania conclusion built from an unverified exit, hidden failures, or a changed rule, and repeat ROU verification after a material configuration change.