2.1.2

The controller image now uses Go 1.26.5.

2.1.1

Release notes:

• The Helm chart uses the latest version of the parallel server proxy (v1.0.1).
• The controller image uses the latest version of the mjssetup library (v1.4.0)

2.1.0

New features:

• The MATLAB Job Scheduler worker and job manager images now install MATLAB at the same location as the official mathworks/matlab image (/opt/matlab/<releaseNumber>, where releaseNumber is the MATLAB release number prefixed with "R", for example R2026a). The old images with MATLAB installed at /opt/matlab will continue to work. You can also now specify a custom MATLAB install location with the matlabRoot parameter.
• The Helm chart now fails to install if any of the PersistentVolumeClaims you have specified in your values.yaml file do not exist.
• The Helm chart now fails to install if any of the older MATLAB releases specified in the additionalSupportedReleases parameter is newer than the release specified in the matlabRelease parameter.
• The Helm chart now fails to install for a cluster with security level 2 if an admin password secret has not been created. Previously, the chart is installed correctly but the controller pod fails to start.

2.0.1

Bug fixes:

Fixed an issue with clusters that use online licensing.

2.0.0

Release Notes:

New Features:

• MATLAB Job Scheduler in Kubernetes now supports MATLAB R2026a.
• You can now use "helm upgrade" to update your MATLAB Job Scheduler deployment in place without needing to uninstall and reinstall the Helm chart. For instructions, see Upgrade Helm Chart.
• MATLAB Job Scheduler in Kubernetes now uses a single proxy pod for all traffic coming into the cluster without needing additional proxy pods for proxying parallel pool traffic to workers. Clusters that support releases older than R2026a still use multiple proxy pods. For more information on the pods in the Helm chart, see Run MATLAB Parallel Server on Kubernetes on the MathWorks website. When you configure your cluster to support older MATLAB releases, you must now also specify the additional releases that your cluster supports using the additionalSupportedReleases parameter. For instructions on configuring a cluster that supports multiple releases, see Run Multiple MATLAB Parallel Server Releases.
• You can configure the number of GPUs to request for each worker pod using the numGPUs parameter. Your Kubernetes cluster must be configured to use NVIDIA GPUs in order to use this setting. For more information on configuring GPUs on a Kubernetes cluster, see Schedule GPUs on the Kubernetes website.
• You can configure the client IP addresses that have access to your cluster using the loadBalancerSourceRanges parameter.
• When you install the Helm chart, it now displays instructions for how to download the cluster profile.
• The job manager and worker pod specs are now configured as part of the Helm chart, allowing for easier modification.
• The controller image now uses Go 1.26.
• The Requirements section now documents the FlexNet network license manager version required for use with this Helm chart (11.19.7 or later).
• MATLAB Parallel Server can now be deployed automatically on Amazon® Elastic Kubernetes Service™ using the MATLAB Parallel Server on Kubernetes in Amazon Web Services reference architecture.

Bug fixes:

• Fixed an issue with the mjs-ingress-proxy-pod getting stuck in a bad state when the job manager IP address changed.
• Fixed an issue where the job manager and workers could be very slow to start on machines with unlimited file descriptor limits.
• The default job manager memory request was decreased from 4Gi to 2Gi, which is suitable for clusters with up to 1000 workers. For larger clusters, add 1Gi for every additional 1000 workers.

1.7.0

New features:

• MATLAB Job Scheduler in Kubernetes now supports MATLAB R2025b.

1.6.0

New features:

• MATLAB Job Scheduler in Kubernetes now supports MATLAB R2025a.

1.5.0

New features:

• Mount custom persistent volumes onto the worker pods using the additionalWorkerPVCs parameter.
• Specify tolerations for the job manager and worker pods using the jobManagerTolerations and workerTolerations parameters.
• Specify custom annotations for the load balancer service using the loadBalancerAnnotations parameter.
• Configure resource requests and limits for all pods deployed by the Helm chart, including the controller and ingress proxy pods.
• The MATLAB Job Scheduler controller now uses Go v1.23.5.

Bug fixes:

• Job manager pods no longer fail to restart after a previous job manager pod stops.
• You can now use the Helm chart as a subchart, as the restrictions in its JSON schema have been removed.

1.4.0

Release Notes:

New features:

• Specify node selectors for the MJS job manager and worker pods using the jobManagerNodeSelector and workerNodeSelector parameters. For more information about node selectors, see Assigning Pods to Nodes on the Kubernetes website.

Bug fixes:

• Workers in R2024b now consistently return licenses to the network license manager.
• The job manager pod now correctly avoids mounting MATLAB from a persistent volume when the jobManagerUsesPVC parameter is set to false.

1.3.0

New features:

• Set up the MATLAB Job Scheduler Helm chart to export cluster monitoring metrics such as cluster status, worker utilization, and licenses in use. Use these metrics to monitor the health of the MATLAB Job Scheduler cluster, diagnose issues, and optimize performance. You can gather the exported metrics with a cluster monitoring system such as Prometheus® and visualize them in a preconfigured Grafana® dashboard. This setup allows for live cluster monitoring and alerts. For more information, see Configure Cluster Monitoring Metrics.
• The Helm chart includes a JSON schema to define the structure of the Helm values file.