> On Sept. 2, 2017, 3:09 a.m., Jie Yu wrote: > > 3rdparty/libprocess/src/libevent_ssl_socket.cpp > > Line 365 (original), 380 (patched) > > <https://reviews.apache.org/r/62051/diff/1/?file=1813748#file1813748line389> > > > > it's also possible that SSL session has already been shutdown. Any > > subsequent `send` call will trigger `BEV_EVENT_ERROR` with `errno=0`.
Good find, updated to reflect this case. - Benjamin ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/62051/#review184412 ----------------------------------------------------------- On Sept. 2, 2017, 12:38 a.m., Benjamin Mahler wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/62051/ > ----------------------------------------------------------- > > (Updated Sept. 2, 2017, 12:38 a.m.) > > > Review request for mesos, Chun-Hung Hsiao, Greg Mann, Jie Yu, and Joseph Wu. > > > Repository: mesos > > > Description > ------- > > This updates the comments to include my findings after having looked > into using BEV_EVENT_READING / BEV_EVENT_WRITING to handle read and > write path errors separately. Unfortunately, we cannot do this using > libevent 2.0.x (see the comments). > > This also explains the "dirty" SSL shutdown case. > > > Diffs > ----- > > 3rdparty/libprocess/src/libevent_ssl_socket.cpp > 0fa7565d1dc49e53890708df154c4d8ea278ad37 > > > Diff: https://reviews.apache.org/r/62051/diff/1/ > > > Testing > ------- > > N/A > > > Thanks, > > Benjamin Mahler > >
