check / check (push) Waiting to run
A new case fails on the current code: the app sends the first part of its body at once and then stalls. The visitor gets that part, and the connection stays open past SWRP_MAX_RESPONSE_TIME_MS. That is the failure seen at random on next, where a busy host let the first part through. The app in the two older cases sent its body after a sleep timed against the limit, so a busy host could let it through too. The app now sends its late parts only once rproxy has ended the request to it, which happens when rproxy cuts off the answer. Those two cases fail on the current code as well. The limit is 2 seconds, so that a busy host still passes on a part sent at once. Model: opus-5-5