There are two kinds of timeouts involved. One is the various *connect_attempts_timeout. This timeout only applies while the connection to the origin is being established (e.g. TLS handshake or TCP three-way for non-TLS). Then the regular proxy.config.http.transaction_no_activity_timeout_out applies while waiting for the first byte and all other bytes exchanged with the origin.
But you are running 71. The connect_timeouts may have been applied until the first byte was received. I think this was fixed to be the proper connect timeout instead of TTFB timeout in the 8.x timeframe. So in that case any POST/PUT method requests would by default wait 1800 seconds for the first byte to return. All other methods would wait the time specified in proxy.config.http.connect_attempts_timeout which appears to default to 30 seconds. Susan On Fri, Nov 20, 2020 at 8:16 AM Andreas Wederbrand <[email protected]> wrote: > Hi! > > I'm running ATS 7.1 as a reverse proxy and one of our origins takes 30+ > seconds occasionally. I thought that > *proxy.config.http.post_connect_attempts_timeout* would control that. It > is set to it's default of 1800 seconds (30 minutes!?) but we still see a > 504 after 30 seconds. > > Am I missing something? Am I looking at the wrong configuration? > > Thanks. >
