https://bz.apache.org/bugzilla/show_bug.cgi?id=61616
--- Comment #9 from Yann Ylavic <ylavic....@gmail.com> --- Created attachment 35427 --> https://bz.apache.org/bugzilla/attachment.cgi?id=35427&action=edit File read buffer and flush threshold tuning Thanks for testing Carsten. I think I see what happens now, the issue is that whenever send_brigade_blocking() is entered it won't return before all the data are flushed, regardless of the remaining threshold. This new patch is mostly unrelated (was mainly written some times ago to allow for better control of the outgoing traffic), but the changes in the core output filtering could address this case (see should_block() and how it's (re)used to bail out without blocking while we remain under the flush max threshold). I didn't committed this patch to trunk yet because it was made for 2.4.x, and some parts quite differ in trunk which makes the frontport non-trivial. Could you please test it and see if it helps for this case too? It should be applied in addition to attachment 35426 (v3) because proxy_connect needs to do the right thing regarding POLLOUT still. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org For additional commands, e-mail: bugs-h...@httpd.apache.org