----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/49280/#review139685 -----------------------------------------------------------
Looks good to me modulo comments. 3rdparty/libprocess/src/libevent_ssl_socket.cpp (lines 808 - 816) <https://reviews.apache.org/r/49280/#comment204987> There seems to also be a LEV_OPT_CLOSE_ON_EXEC to support the accept4 SOCK_CLOEXEC flag, but it appears it only made it into the 2.1.5 beta: https://github.com/libevent/libevent/commit/4970329a88e3e8ea10317478a642b99d4b28a283 Worth adding a note that this would be how to do it atomically if libevent and the OS supports it. 3rdparty/libprocess/src/libevent_ssl_socket.cpp (line 815) <https://reviews.apache.org/r/49280/#comment204999> Can you put a Failure Future<Socket> into the accept queue? (Much like how the poll-based socket works?) That way we can deal with the error in the accept loop in libprocess. - Benjamin Mahler On June 27, 2016, 7:38 p.m., Joseph Wu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/49280/ > ----------------------------------------------------------- > > (Updated June 27, 2016, 7:38 p.m.) > > > Review request for mesos, Avinash sridharan, Benjamin Mahler, Artem > Harutyunyan, and Joris Van Remoortere. > > > Bugs: MESOS-5723 > https://issues.apache.org/jira/browse/MESOS-5723 > > > Repository: mesos > > > Description > ------- > > Incoming sockets are leaked when the agent forks because incoming > sockets are not modified with the CLOEXEC option. > > > Diffs > ----- > > 3rdparty/libprocess/src/libevent_ssl_socket.cpp > 3f80e863fdbb62d4cf7ce1b46c468a09f0694f56 > > Diff: https://reviews.apache.org/r/49280/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Joseph Wu > >
