MATLAB: Does MATLAB Parallel Server using HPC validation fail when validating on step 2 job test

MATLAB Parallel Server

I have an MATLAB Parallel Server cluster using Microsoft HPC as the scheduler and when I try to validate, validation fails on step 2 with an error similar to this one:

Unable to find log file

"D:\Temp\MATLAB\Job1\Task1.log".

Getting data for HPC Server 2008 JobID 1401.

Error Message : Task 1401.1 failed. Please check the failed task for more

details on the failure.

Getting data for HPC Server 2008 JobID 1401 and TaskID 1.

CommandLine : "D:\Program Files\MATLAB\R2014a\bin\worker.bat"

Stdout sent : D:\Temp\MATLAB\Job1\Task1.log

Error Message : Unable to open standard output file on node

HPC-SERVER:Exception 'Failed to create standard output file

'D:\Temp\MATLAB\Job1\Task1.log', The system cannot find

the path specified' reported creating the task.

Best Answer

This error indicates that the job storage location defined in your cluster profile is not a directory which the user running MATLAB Parallel Server on the HPC server can write to. This can happen because:

The directory does not exist
The directory is not reachable from the HPC server head node
The directory exists but the user that is running MATLAB Parallel Server does not have permission to write to that directory

To resolve this issue ensure that the directory you have specified in your cluster profile can be written to by the user running MATLAB Parallel Server on the head node.

Related Solutions

MATLAB: Do I receive an error when submitting jobs to HPC Server from a 32-bit Windows client

32-bit clients are supported for HPC starting with R2011a. For previous releases, see the winformation below.

Parallel Computing Toolbox only supports job submission to an HPC Server 2008 cluster from 64-bit Windows clients. In order to submit to this type of cluster, it must be done from a 64-bit Windows client. It is possible to submit from a 32-bit client, but this must be done by setting the "ClusterVersion" of the configuration to ccs. This allows the jobs to submit using the old CCS mode, which does not use the advanced HPC functions.

NOTE: Starting in R2019a the following name changes occurred:
MATLAB Distributed Computing Server was renamed to MATLAB Parallel Server
mdce_def was renamed to mjs_def
mdce binary was renamed to mjs

MATLAB: Am I getting warning messages when submitting a BATCH job to a cluster running MATLAB Distributed Computing Server 5.0 (R2010b)

This warning occurs because prior to evaluating the script, BATCH tries to change the worker's working directory to the same directory as the submitting MATLAB.

When the worker is on a different machine than the client, and the file system is not shared between the client and worker, this is not possible, and MATLAB gives the warning messages.

To work around this issue set 'CurrentDirectory' parameter in BATCH to '.', as follows:

job = batch('myScript','CaptureDiary',true, 'CurrentDirectory', '.'); 
waitForState(job) 
diary(job) 
destroy(job)

With 'CurrentDirectory' parameter set to '.' the BATCH command will not attempt to change the working directory on the worker.

Best Answer

Related Solutions

MATLAB: Do I receive an error when submitting jobs to HPC Server from a 32-bit Windows client

MATLAB: Am I getting warning messages when submitting a BATCH job to a cluster running MATLAB Distributed Computing Server 5.0 (R2010b)

Related Question