hadoop yarn 运行分布式的应用时出现载入不了库文件错误

我在hadoop 2.7.1版本 运行https://github.com/alibaba/mpich2-yarn.git，来跑我的一个分布式应用
运行时出现如下错误：
17/03/08 04:30:27 INFO client.Client: Initializing ApplicationMaster
17/03/08 04:30:35 INFO client.Client: all containers are launched successfully, executing mpiexec...
17/03/08 04:30:35 INFO client.Client: [stderr] /home/ubuntu/hdfstmp/mpiexecs/appattempt_1488941680484_0003_000001/MPIExec: error while loading shared libraries: libjvm.so: cannot open shared object file: No such file or directory
17/03/08 04:30:35 INFO client.Client: [stderr] /home/ubuntu/hdfstmp/mpiexecs/appattempt_1488941680484_0003_000001/MPIExec: error while loading shared libraries: libjvm.so: cannot open shared object file: No such file or directory
17/03/08 04:30:35 INFO client.Client: [stdout]
17/03/08 04:30:35 INFO client.Client: [stdout] ===================================================================================
17/03/08 04:30:35 INFO client.Client: [stdout] = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
17/03/08 04:30:35 INFO client.Client: [stdout] = EXIT CODE: 127
17/03/08 04:30:35 INFO client.Client: [stdout] = CLEANING UP REMAINING PROCESSES
17/03/08 04:30:35 INFO client.Client: [stdout] = YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
17/03/08 04:30:35 INFO client.Client: [stdout] ===================================================================================

个人认为应该是没有将应用应用运行所需的资源提交给RM，如果真是这样，我需要怎么提交呢，望各位大神给点建议。。。