parfor loop error on Azure VM

I got the following error message whenever I tried to use parallel computing using parfor. I have tried very simple parfor routine, thus the error is not related to the parfor loop.

parfor i=1:100

a = i;

end

Starting parallel pool (parpool) using the 'local' profile ...

Warning: Cannot cancel or destroy a job that was not created by this Local cluster.

> In parallel.cluster.Local/cancelOrDestroyJob (line 34)

In parallel.cluster/Local/hCancelJob (line 294)

In parallel.internal.cluster/CJSJobMethods/cancelOneJob (line 51)

In parallel.job.CJSConcurrentJob>@(job,ce)CJSJobMethods.cancelOneJob(job.Parent,job,job.Support,ce) (line 57)

In parallel.job/CJSConcurrentJob/cancelJob (line 57)

In parallel.Job>iCancelJobs (line 1558)

In parallel.Job>@(jobs)iCancelJobs(jobs,exceptionBuilder) (line 1367)

In parallel.internal.cluster.hetfun (line 57)

In parallel/Job/cancel (line 1367)

In parallel/Cluster/hDeleteOneJob (line 1014)

In parallel.internal.pool.InteractiveClient>iDeleteJobs (line 877)

In parallel.internal.pool/InteractiveClient/pStopLabsAndDisconnect (line 585)

In parallel.internal.pool.InteractiveClient>iCleanupIfStartupFailed (line 951)

In parallel.internal.pool.InteractiveClient>@()iCleanupIfStartupFailed(obj) (line 283)

In onCleanup/delete (line 80)

In parallel.internal.pool/InteractiveClient/start (line 226)

In parallel.internal.pool.AbstractClusterPool>iStartClient (line 826)

In parallel.internal.pool/AbstractClusterPool/hBuildPool (line 596)

In parallel.internal.pool.doParpool (line 20)

In parpool (line 122)

In parallel.internal.pool/PoolArrayManager/getOrAutoCreateWithCleanup (line 58)

In pctTryCreatePoolIfNecessary (line 28)

In parallel_function (line 418)

In untitled2 (line 5)

Warning: Cannot cancel or destroy a job that was not created by this Local cluster.

> In parallel.cluster.Local/cancelOrDestroyJob (line 34)

In parallel.cluster/Local/hDestroyJob (line 298)

In parallel.internal.cluster/CJSJobMethods/destroyOneJob (line 71)

In parallel.job.CJSConcurrentJob>@(job)CJSJobMethods.destroyOneJob(job.Parent,job,job.Support,job.SupportID) (line 52)

In parallel.job/CJSConcurrentJob/destroyJob (line 52)

In parallel.Job>iDeleteJobs (line 1533)

In parallel.internal.cluster.hetfun (line 57)

In parallel/Job/delete (line 1313)

In parallel/Cluster/hDeleteOneJob (line 1030)

In parallel.internal.pool.InteractiveClient>iDeleteJobs (line 877)

In parallel.internal.pool/InteractiveClient/pStopLabsAndDisconnect (line 585)

In parallel.internal.pool.InteractiveClient>iCleanupIfStartupFailed (line 951)

In parallel.internal.pool.InteractiveClient>@()iCleanupIfStartupFailed(obj) (line 283)

In onCleanup/delete (line 80)

In parallel.internal.pool/InteractiveClient/start (line 226)

In parallel.internal.pool.AbstractClusterPool>iStartClient (line 826)

In parallel.internal.pool/AbstractClusterPool/hBuildPool (line 596)

In parallel.internal.pool.doParpool (line 20)

In parpool (line 122)

In parallel.internal.pool/PoolArrayManager/getOrAutoCreateWithCleanup (line 58)

In pctTryCreatePoolIfNecessary (line 28)

In parallel_function (line 418)

In untitled2 (line 5)

Mismatch between number of environment names and values

[Edric Ellis]

I think this is some sort of configuration problem. I suggest contacting MathWorks support directly for help with this.

[Jose Cortes-Briones]

I have the same issue. I'm running Matlab 2022a using the docker for NVIDIA systems. Did you ever get an answer?

Answer 1

[Raghav Bansal]

Hi Chamna,

The log suggests that there is some issue in credential handling. Here are few points which can help you to resolve the same:

1. Please ensure there is only one version of MATLAB installed on your machine.
2. Try logging out and logging in again in the MATLAB.

If the above does not work, you may try to perform a clean installation of the MATLAB. You can refer to the below MATLAB answer to know more about the same:

Clean installation of MATLAB

The error "Mismatch between number of environment names and values" should also get resolved by this. You can use the following commands in MATLAB command window to confirm this:

dbstop('in', fullfile(matlabroot, 'toolbox/parallel/cluster/+parallel/+cluster/@Local/hSubmitCommunicatingJob.m'), 'at', '96')
c = parcluster('local'); %Name of the cluster profile
cd(c.JobStorageLocation)
setenv MDCE_DEBUG true
j = c.createCommunicatingJob();
j.createTask(@pwd, 1, {});
j.submit
cd(j.Name)
j.wait

% MATLAB will enter in the debugging mode

K>>envNames
K>>envValues

The number of values should be same for both 'envNames' and the 'envValues'.

Hope it helps!

Regards,

Raghav