----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/27865/#review62279 -----------------------------------------------------------
This is great! The continuation of the Libproces-From work from before. ;-) 3rdparty/libprocess/src/encoder.hpp <https://reviews.apache.org/r/27865/#comment104302> s/in_addr */in_addr*/ And double check the place you copied the code from to fix it too. ;-) 3rdparty/libprocess/src/encoder.hpp <https://reviews.apache.org/r/27865/#comment104301> If we can use 'Libprocess-From' that would be preferred since we've already introduced it (but if we can't because of HTTP restrictions on header field-value let me know). - Benjamin Hindman On Nov. 13, 2014, 8:27 p.m., Dario Rexin wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/27865/ > ----------------------------------------------------------- > > (Updated Nov. 13, 2014, 8:27 p.m.) > > > Review request for mesos and Benjamin Hindman. > > > Bugs: MESOS-2071 > https://issues.apache.org/jira/browse/MESOS-2071 > > > Repository: mesos-git > > > Description > ------- > > While working on a pure Scala driver for Mesos, I discovered that libprocess > does not generate valid HTTP. It uses chunked encoding combined with HTTP > 1.0, which has only been added in HTTP 1.1. Additionally it stores the PID in > the User-Agent field with an incompatibe format ('@' and ':' are not > allowed). This patch sets the HTTP version to 1.1, adds an empty Host header > (Host is mandatory in 1.1) and adds 3 new fields: X-Mesos-Id, X-Mesos-Ip, > X-Mesos-Port containing the PID parts. The User-Agent header is left > untouched for backwards compatibility reasons. Even strict HTTP parser > shouldn't error, but just ignore everything between the invalid character and > the CRLF. > > > Diffs > ----- > > 3rdparty/libprocess/src/encoder.hpp 9c5aa81 > 3rdparty/libprocess/src/tests/process_tests.cpp b985fb7 > > Diff: https://reviews.apache.org/r/27865/diff/ > > > Testing > ------- > > - ran `make check` on Linux and OSX > - setup a cluster consisting of 1 Master, 1 Slave and 1 Marathon instance, > deployed a couple of tasks and destroyed them afterwards > > > Thanks, > > Dario Rexin > >
