-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/47207/
-----------------------------------------------------------

Review request for mesos, Alexander Rukletsov, Ben Mahler, haosdent huang, and 
Joris Van Remoortere.


Bugs: MESOS-5340
    https://issues.apache.org/jira/browse/MESOS-5340


Repository: mesos


Description
-------

The approach wraps the Socket::accept() returned future so that
libprocess will immediately be able to trigger a new accept,
allowing new connections to get established without awaiting
a full connection validation of that initial accept-call.


Diffs
-----

  3rdparty/libprocess/include/process/socket.hpp 
4cb49680d1304899a4ee675ac07379e51d9c55b1 
  3rdparty/libprocess/src/libevent_ssl_socket.hpp 
e773fadeb31ca384264a425bdc8d093804e45a82 
  3rdparty/libprocess/src/libevent_ssl_socket.cpp 
b829e7dd2c0e7730ac1579c042f79f40666b19b9 
  3rdparty/libprocess/src/poll_socket.hpp 
d04f3f2d1bcf70464ac659b29f96574bbd233414 
  3rdparty/libprocess/src/poll_socket.cpp 
e68c7836b6a79253fa646c1919d0f331f21ec131 
  3rdparty/libprocess/src/process.cpp dcaa64633d1eea330649c563635642928164d73c 

Diff: https://reviews.apache.org/r/47207/diff/


Testing
-------

WIP!

No make check was even attempted - tests will not compile. This patch was 
initially developed by AlexR and is meant as a starting point of further 
discussions. Functional testing using the scenario described in the attached 
JIRA was succesful, the idle connection does not prevent new connection 
processing anymore.


Thanks,

Till Toenshoff

Reply via email to