I followed the myriad setup instructions , and was able to get resource manager invoke the myriad scheduler and talk to the mesos master. But I see the following error in the mesos slave logs and my yarn submissions are stuck.
My setup is as follows: 1. Hadoop 2.7.1 2. Jdk8 3. Mesos Version: 0.25.0 4. 1 master + 2 slaves 5. ubuntu 14.04 + Kernel Linux master.dev 3.19.0-33-generic #38~14.04.1-Ubuntu SMP Fri Nov 6 18:17:28 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux Given this team is running with this, it is hard for me to presume this is a argument overflow issue and would require somekind of a kernel recompile : http://www.linuxjournal.com/article/6060?page=0,0. I am also thinking if to recompile mesos for better diagnostics. the subprocess.cpp seems to have better logging in master : https://github.com/apache/mesos/blob/master/3rdparty/libprocess/src/subprocess.cpp than in 0.25.0 ABORT: (/tmp/mesos-build/mesos-repo/3rdparty/libprocess/src/subprocess.cpp:177): Failed to os::execvpe in childMain: Argument list too long*** Aborted at 1449220361 (unix time) try "date -d @1449220361" if you are using GNU date *** PC: @ 0x7fbfd2c66cc9 (unknown) *** SIGABRT (@0x231d) received by PID 8989 (TID 0x7fbfc944a700) from PID 8989; stack trace: *** @ 0x7fbfd3005340 (unknown) @ 0x7fbfd2c66cc9 (unknown) @ 0x7fbfd2c6a0d8 (unknown) @ 0x40a902 _Abort() @ 0x40a93c _Abort() @ 0x7fbfd477ac3b process::childMain() @ 0x7fbfd477cc6d std::_Function_handler<>::_M_invoke() @ 0x7fbfd2d2a47d (unknown)