MATLAB: MDCE Causing Blue Screen of Death (Clock Watchdog Timeout)

batchbluecomputingdeathdistributedenvironmentMATLABMATLAB Parallel Servermdceparallelscreensimulinkwatchdog

We have a 24 core computer that has a MDCE license on it. Currently I have 23 workers set for the machine. I am able to successfuly submit complex simulink runs to the machine and retrieve the data when they finish. However, once I start submitting enough jobs where some are in the queue the machine crashes given a BSoD error of "Clock Watchdog Timeout". I am not entirely sure how many jobs causes the error. Once I had 24 jobs going and they all finished without a problem.

Best Answer

Solved. In the "admincenter" turn off automatic updates. No more "Clock Watchdog Timeout" blue screen of death errors.

Related Solutions

MATLAB: Is the communicating job stuck behind a concurrent job in queue in R2017b

The resolution is to avoid starting a parallel pool when submitting other jobs to the cluster.

The "parpool" object is generally used for interactive "parfor" or "smpd" operations. When the parallel pool is created with "parpool", it creates a concurrent job on the cluster that readies all of the workers to be available for when the user interactively calls "parfor" or "smpd". The concurrent job will continue running, causing other jobs to wait in queue until the parallel pool is shut down.

MATLAB: How to manually validate the MATLAB cluster

When validation fails the feedback that is displayed is often not very descriptive. If you run a manual validation job while in verbose mode you might be able to obtain a more specific error message. Please see the following instructions on how to perform a manual validation.

NOTE: The following commands work whether you are validating a local profile or an MATLAB Parallel Server profile. Please make sure you are validating the correct profile before proceeding.

In order to place MATLAB in verbose mode run the following command within the MATLAB command Window.

setSchedulerMessageHandler(@disp)

Then run each of the following jobs:\n

Independent Job

cluster = parcluster();
job = createJob (cluster);
createTask(job, @sum, 1, {[1 1]});
submit(job);
wait(job);
out = fetchOutputs(job)

\n \n

SPMD Job

cluster = parcluster();
job = createCommunicatingJob(cluster, 'Type', 'spmd') ;
createTask(job, @labindex, 1, {});
submit(job);
wait(job);
out = fetchOutputs(job)

Starting with MATLAB R2013b to run a MATLAB interactive job, use the "parpool" command:

Parpool

parpool('name_of_profile', 2)

\n \nIf using MATLAB R2013a or older to run a MATLAB interactive job, use the "matlabpool" command:\n

Matlabpool

matlabpool('name_of_profile', 2)

If any of these jobs fail you should be presented with a more descriptive error message.

NOTE: Starting in R2019a the following name changes occurred:\n

MATLAB Distributed Computing Server was renamed to MATLAB Parallel Server
"mdce_def" was renamed to "mjs_def"
"mdce" binary was renamed to "mjs"

Best Answer

Related Solutions

MATLAB: Is the communicating job stuck behind a concurrent job in queue in R2017b

MATLAB: How to manually validate the MATLAB cluster

Related Question