Choose Solution to Install and Configure MATLAB Parallel Server

Consult the following tables for help with the most common tasks while setting up MATLAB^® Parallel Server™. The tables help you to choose when you want to:

• Activate the License for, Install, and Configure MATLAB Parallel Server on the Cluster

• Configure a MATLAB Client for a User to Submit Jobs to the Cluster

• Upgrade MATLAB Parallel Server

Activate the License for, Install, and Configure MATLAB Parallel Server on the Cluster

Consult the following table if you are getting started and you want to install and configure MATLAB Parallel Server for your cluster.

Scenario		Solution
You have a Slurm scheduler in your cluster.		Follow Install MATLAB Parallel Server for Slurm.
You have an existing scheduler in your cluster, such as LSF® PBS Torque Grid Engine Microsoft® HPC Pack HTCondor Hadoop® Spark™ All other third-party schedulers		Follow Install MATLAB Parallel Server for Other Third-Party Schedulers.
You do not have an existing scheduler in your cluster.	You want to manage licensing of your cluster users on premises.	Set up the MATLAB Job Scheduler, which comes with MATLAB Parallel Server. Follow Install for MATLAB Job Scheduler with Network License Manager.
	You want to use MathWorks® online licensing for your cluster users.	Set up the MATLAB Job Scheduler, which comes with MATLAB Parallel Server. Follow Install for MATLAB Job Scheduler with Online Licensing.

Configure a MATLAB Client for a User to Submit Jobs to the Cluster

Consult the following table if you already have a cluster configured and you want to configure a new MATLAB client to submit jobs to the cluster.

Scenario	Solution
The cluster uses a third-party scheduler and the scheduler has out-of-the-box support in MATLAB. Schedulers with out-of-the-box support are: Microsoft HPC Pack and Windows® HPC Server Slurm LSF Torque PBS Pro® and OpenPBS Grid Engine HTCondor	See Install Software on Local Desktop. Then, Slurm, LSF, PBS Pro, OpenPBS, Grid Engine, Torque – If the client is a Windows machine, start from Configure Windows Firewalls on Server. Otherwise, start from Create Cluster Profile and Validate Installation. HPC Server – Start from Configure Client Computer for HPC Pack.
The cluster is a Spark cluster.	Start from Configure for Spark Clusters
The cluster is a Hadoop cluster.	Start from Configure a Hadoop Cluster.
The cluster uses a third-party scheduler and the scheduler does not have out-of-the-box support in MATLAB.	Follow Configure Using the Generic Scheduler Interface.
The cluster uses the MATLAB Job Scheduler.	If the cluster is using MathWorks online licensing, ask your cluster administrator to add the new user. For more information, see Add Licensed End Users. Then, follow Connect the MATLAB Client to the MATLAB Parallel Server Cluster. Otherwise, follow Connect MATLAB Client to MATLAB Parallel Server Cluster.

Note

You can export a cluster profile and import it in another client machine. For more information, see Import and Export Cluster Profiles (Parallel Computing Toolbox). This is useful when you are configuring another client machine with the same configuration, to avoid recreating the cluster profile manually.

Upgrade MATLAB Parallel Server

Consult the following table if you want to configure a new release of MATLAB Parallel Server.

Scenario	Solution
The cluster uses a third-party scheduler.	Follow Install MATLAB Parallel Server for Other Third-Party Schedulers to Activate and install the new version of MATLAB Parallel Server on the cluster. Configure the cluster and client machines.
The cluster uses the MATLAB Job Scheduler.	Stop the mjs service of the older release. For instructions, see Stop mjs Services of Old Installation. Install the new version of MATLAB Parallel Server on the cluster and configure the MATLAB Job Scheduler. For more information, see Run Multiple MATLAB Parallel Server Versions. For help with activation, installation, and configuration, consult the table in Activate the License for, Install, and Configure MATLAB Parallel Server on the Cluster.