We have an installation of MATLAB that uses the MPD build of the MPICH2 library. The command "matlabpool open local 4" generates following error messages on our cluster:
>>matlabpool open local 4 Starting matlabpool using the parallel configuration 'local'. Waiting for parallel job to start... Performing parallel job cleanup... Done. ??? Error using ==> distcomp.interactiveclient.pGetSockets>iThrowIfBadParallelJobStatus at 107 The interactive parallel job finished without any messages. Error in ==> matlabpool at 90 client.start('matlabpool', numlabs, config, 'nogui');
More information from troubleshooting steps is shown below:
setSchedulerMessageHandler(@disp)matlabpool open local 4 Starting matlabpool using the parallel configuration 'local'. Waiting for parallel job to start... Checking parallel job status. Checking parallel job status. Checking parallel job status. Checking parallel job status. [...] Job state is 'failed'
The following script was tried:
restoredefaultpathsetenv('MDCE_DEBUG', 'true');sched = findResource('scheduler', 'Type', 'local');job = sched.createParallelJob('Max', 4);job.createTask(@pwd, 1 );job.submit;disp('about to wait for job...');job.wait;sched.getDebugLog(job)
… and the output of above was:
MATLAB core dump: Exit on fatal error (no core) enabled. Warning: We're about to use the 1.0.3-mpd build > In mpiLibConf at 3 In distcomp_evaluate_filetask>iSetCurrentTaskFromEnvironment at 254 In distcomp_evaluate_filetask at 18 In decodeLocalSingleTask In decodeLocalSingleTask
Moreover, we have only one copy of "pathdef.m" on the MATLAB path, and it has all the necessary paths required to run the toolbox.
Best Answer