Version 1.5.0

Author: hannahpullen

README.md

@@ -503,6 +503,7 @@ kubectl get secrets mjs-metrics-client-certs --template="{{.data.prometheus.key
 
 MATLAB Job Scheduler in Kubernetes uses a Kubernetes load balancer service to expose MATLAB Job Scheduler to MATLAB clients running outside of the Kubernetes cluster.
 By default, the Helm chart creates the load balancer for you.
+You can customize the annotations on the Kubernetes load balancer service by setting the `loadBalancerAnnotations` parameter in your `values.yaml` file.
 You can also create and customize your own load balancer service before you install the Helm chart.
 
 Create a Kubernetes load balancer service `mjs-ingress-proxy` to expose MATLAB Job Scheduler to MATLAB clients running outside of the Kubernetes cluster.

chart/mjs/Chart.yaml

@@ -3,5 +3,5 @@ apiVersion: v2
 name: mjs
 description: A Helm chart for MATLAB (R) Job Scheduler in Kubernetes
 type: application
-version: 1.4.0
-appVersion: 1.4.0
+version: 1.5.0
+appVersion: 1.5.0

chart/mjs/templates/_paths.tpl

@@ -6,14 +6,20 @@
 /opt/matlab
 {{- end -}}
 
-# Checkpoint base on containers
-# If running as non-root user and not mounting checkpointbase, use a directory that a non-root user can create 
+# Checkpoint base on containers; all MJS data except the job database is stored here.
+# Note that this data is not persisted across pod restarts, as we want a fresh MJS every time.
 {{- define "paths.checkpointbase" -}}
+/tmp/checkpoint
+{{- end -}}
+
+# Database directory on the job manager pod.
+# If running as non-root user and not mounting a checkpoint PVC, use a directory that a non-root user can create 
+{{- define "paths.databasedir" -}}
 {{- $isNonRoot := ne (int $.Values.jobManagerUserID) 0 -}}
 {{- if and $isNonRoot (empty $.Values.checkpointPVC) -}} 
-/tmp/checkpoint
+/tmp/database
 {{- else -}}
-/mjs/checkpoint
+/mjs/database
 {{- end -}}
 {{- end -}}
 

chart/mjs/templates/controller-configmap.yaml

@@ -1,5 +1,5 @@
 # Config file for the MJS controller.
-# Copyright 2024 The MathWorks, Inc.
+# Copyright 2024-2025 The MathWorks, Inc.
 {{- $workerImageTag := .Values.matlabImageTag | default .Values.matlabRelease }}
 {{- $jobManagerImageTag := .Values.jobManagerImageTag | default .Values.matlabRelease }}
 {{- $workerImage := .Values.matlabImage }}
@@ -25,9 +25,10 @@ data:
         {{- $comma = "," }}
       {{- end }}
       ],
+      "AdditionalWorkerPVCs": {{ toJson (.Values.additionalWorkerPVCs | default dict) }},
       "BasePort": {{ .Values.basePort | int }},
       "CertFileName": {{ include "paths.certFile" . | quote }},
-      "CheckpointBase": {{ include "paths.checkpointbase" . | quote }},
+      "CheckpointBase": {{ include "paths.databasedir" . | quote }},
       "CheckpointPVC": {{ .Values.checkpointPVC | quote }},
       "ClusterHost": {{ .Values.clusterHost | quote }},
       "ControllerLogfile": {{ include "paths.controllerLog" . | quote }},
@@ -46,6 +47,7 @@ data:
       "JobManagerMemoryRequest": {{ .Values.jobManagerMemoryRequest | quote }},
       "JobManagerGroupID": {{ .Values.jobManagerGroupID }},
       "JobManagerNodeSelector": {{ toJson .Values.jobManagerNodeSelector }},
+      "JobManagerTolerations": {{ toJson .Values.jobManagerTolerations | quote }},
       "JobManagerUserID": {{ .Values.jobManagerUserID }},
       "JobManagerUsesPVC": {{ .Values.jobManagerUsesPVC }},
       "JobManagerUID": {{ uuidv4 | quote }},
@@ -97,6 +99,7 @@ data:
       "WorkerMemoryRequest": {{ .Values.workerMemoryRequest | quote }},
       "WorkerLogPVC": {{ .Values.workerLogPVC | quote }},
       "WorkerNodeSelector": {{ toJson .Values.workerNodeSelector }},
+      "WorkerTolerations": {{ toJson .Values.workerTolerations | quote }},
       "WorkerPassword": {{ .Values.workerPassword | quote }},
       "WorkersPerPoolProxy": {{ .Values.workersPerPoolProxy }},
       "WorkerUsername": {{ .Values.workerUsername | quote }},

chart/mjs/templates/controller-deployment.yaml

@@ -24,12 +24,30 @@ spec:
 
       # Schedule on same nodes as the job manager
       nodeSelector: {{ toJson .Values.jobManagerNodeSelector }}
+      tolerations: {{ toJson .Values.jobManagerTolerations }}
 
       containers:
       - name: {{ $name }}
         image: {{ printf "%s:%s" .Values.controllerImage $controllerImageTag }}
         imagePullPolicy: {{ .Values.controllerImagePullPolicy }}
 
+        # Set optional resource requests and limits
+        resources:
+          requests:
+            {{- if .Values.controllerCPURequest }}
+            cpu: {{ .Values.controllerCPURequest }}
+            {{- end }}
+            {{- if .Values.controllerMemoryRequest }}
+            memory: {{ .Values.controllerMemoryRequest }}
+            {{- end }}
+          limits:
+            {{- if .Values.controllerCPULimit }}
+            cpu: {{ .Values.controllerCPULimit }}
+            {{- end }}
+            {{- if .Values.controllerMemoryLimit }}
+            memory: {{ .Values.controllerMemoryLimit }}
+            {{- end }}
+
         # The controller process requires the path to a config file as an input argument
         # This file is mounted from a ConfigMap (defined in mjs.yaml)
         {{- $configMapDir := "/config" }}
@@ -57,4 +75,4 @@ spec:
       - name: {{ $logVolName }}
         persistentVolumeClaim:
           claimName: {{ .Values.logPVC }}
-      {{- end }}
+      {{- end }}

chart/mjs/templates/ingress-proxy-deployment.yaml

@@ -22,12 +22,30 @@ spec:
 
       # Schedule on same nodes as the job manager
       nodeSelector: {{ toJson .Values.jobManagerNodeSelector }}
+      tolerations: {{ toJson .Values.jobManagerTolerations }}
 
       containers:
         - name: haproxy
           image: {{ $.Values.haproxyImage }}
           imagePullPolicy: {{ $.Values.haproxyImagePullPolicy }}
 
+          # Set optional resource requests and limits
+          resources:
+            requests:
+              {{- if .Values.haproxyCPURequest }}
+              cpu: {{ .Values.haproxyCPURequest }}
+              {{- end }}
+              {{- if .Values.haproxyMemoryRequest }}
+              memory: {{ .Values.haproxyMemoryRequest }}
+              {{- end }}
+            limits:
+              {{- if .Values.haproxyCPULimit }}
+              cpu: {{ .Values.haproxyCPULimit }}
+              {{- end }}
+              {{- if .Values.haproxyMemoryLimit }}
+              memory: {{ .Values.haproxyMemoryLimit }}
+              {{- end }}
+
           # Pass the config file path as an input argument
           {{- $configDir := "/usr/local/etc/haproxy/" }}
           args:

chart/mjs/templates/ingress-proxy-service.yaml

@@ -8,6 +8,7 @@ metadata:
   name: {{ $name }}
   labels:
     app: {{ $name }}
+  annotations: {{ toJson .Values.loadBalancerAnnotations }}
 spec:
   type: LoadBalancer
   selector:

chart/mjs/templates/mjs-configmap.yaml

@@ -51,20 +51,20 @@ data:
     # Add LDAP certificate to the Java trust store
     chmod +w {{ $trustStorePath }}
     yes | {{ include "paths.matlabroot" . }}/sys/java/jre/glnxa64/jre/bin/keytool -import -keystore {{ $trustStorePath }} -file {{ include "paths.ldapCert" . }} -storepass changeit > /dev/null
- 
+
     {{ end -}}
     # Start the MJS service
     {{ printf "%s/mjs" $binDir }} start -mjsdef {{ $localMJSDef }} -loglevel {{ .Values.logLevel }} -disablereliableconnections
 
     # Start the job manager
-    {{- $jmCommand := printf "%s/startjobmanager -name \"%s\" -baseport %d" $binDir .Values.jobManagerName $basePort }}
+    {{- $jmCommand := printf "%s/startjobmanager -name \"%s\" -baseport %d -databaseDirectory %s" $binDir .Values.jobManagerName $basePort (include "paths.databasedir" .) }}
     {{- if .Values.requireClientCertificate }}
       {{- $jmCommand = printf "%s -certificate \"%s\"" $jmCommand (include "paths.certPath" .) }}
     {{- end }}
     {{- if .Values.requireScriptVerification }}
       {{- $jmCommand = printf "%s -secretfile \"%s\"" $jmCommand (include "paths.secretPath" .) }}
     {{- end }}
-    {{ $jmCommand }} || echo "startjobmanager failed; there may already be a job manager running"
+    {{ $jmCommand }}
 
     # Keep the container running
     sleep infinity

chart/mjs/values.schema.json

@@ -7,6 +7,12 @@
                 "type": "string"
             }
         },
+        "additionalWorkerPVCs": {
+            "type": "object",
+            "additionalProperties": {
+                "type": "string"
+            }
+        },
         "adminUser": {
             "type": "string"
         },
@@ -28,6 +34,18 @@
         "clusterHost": {
             "type": "string"
         },
+        "controllerCPULimit": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
+        "controllerCPURequest": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
         "controllerImage": {
             "type": "string"
         },
@@ -38,6 +56,18 @@
         "controllerImageTag": {
             "type": "string"
         },
+        "controllerMemoryLimit": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
+        "controllerMemoryRequest": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
         "enableServiceLinks": {
             "type": "boolean"
         },
@@ -50,13 +80,40 @@
                 "type": "string"
             }
         },
+        "global": {
+            "type": "object"
+        },
+        "haproxyCPULimit": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
+        "haproxyCPURequest": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
         "haproxyImage": {
             "type": "string"
         },
         "haproxyImagePullPolicy": {
             "type": "string",
             "enum": ["Always", "Never", "IfNotPresent"]
         },
+        "haproxyMemoryLimit": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
+        "haproxyMemoryRequest": {
+            "anyOf": [
+            { "type": "string" },
+            { "type": "number" }
+            ]
+        },
         "idleStop": {
             "type": "integer",
             "minimum": 0
@@ -111,6 +168,15 @@
                 "type": "string"
             }
         },
+        "jobManagerTolerations": {
+            "type": "array",
+            "items": {
+                "type": "object",
+                "additionalProperties": {
+                    "type": "string"
+                }
+            }
+        },
         "jobManagerUserID": {
             "type": "integer",
             "minimum": 0
@@ -140,6 +206,12 @@
             "type": "integer",
             "minimum": 1
         },
+        "loadBalancerAnnotations": {
+            "type": "object",
+            "additionalProperties": {
+                "type": "string"
+            }
+        },
         "logLevel": {
             "type": "integer",
             "minimum": 0,
@@ -308,6 +380,15 @@
             "type": "integer",
             "minimum": 1
         },
+        "workerTolerations": {
+            "type": "array",
+            "items": {
+                "type": "object",
+                "additionalProperties": {
+                    "type": "string"
+                }
+            }
+        },
         "workerUsername": {
             "type": "string"
         }

chart/mjs/values.yaml

@@ -1,5 +1,5 @@
 # Default values for MATLAB Job Scheduler (MJS) in Kubernetes.
-# Copyright 2024 The MathWorks, Inc.
+# Copyright 2024-2025 The MathWorks, Inc.
 
 # Release number of the MATLAB version to use
 matlabRelease: "r2024a"
@@ -60,10 +60,20 @@ poolProxyCPULimit: "" # CPU limit for each parallel pool proxy process
 poolProxyCPURequest: "0.5" # CPU request for each parallel pool proxy process
 poolProxyMemoryLimit: "" # Memory limit for each parallel pool proxy process
 poolProxyMemoryRequest: "500Mi" # Memory request for each parallel pool proxy process
+controllerCPULimit: "" # CPU limit for the MJS controller
+controllerCPURequest: "100m" # CPU request for the MJS controller
+controllerMemoryLimit: "" # Memory limit for the MJS controller
+controllerMemoryRequest: "128Mi" # Memory request for the MJS controller
+haproxyCPULimit: "" # CPU limit for the HAProxy pod
+haproxyCPURequest: "100m" # CPU request for the HAProxy pod
+haproxyMemoryLimit: "" # Memory limit for the HAProxy pod
+haproxyMemoryRequest: "256Mi" # Memory request for the HAProxy pod
 
 # Node settings
 jobManagerNodeSelector: {} # Node selector for the job manager, specified as key-value pairs
 workerNodeSelector: {} # Node selector for the workers, specified as key-value pairs
+jobManagerTolerations: [] # Tolerations for the job manager pod
+workerTolerations: [] # Tolerations for the worker pods
 
 # Auto-scaling settings
 idleStop: 300 # Time after which idle worker pods will be removed
@@ -72,6 +82,7 @@ stopWorkerGracePeriod: 60 # Grace period in seconds for running stopworker
 
 # Network settings
 autoCreateLoadBalancer: true # Flag to automatically create a Kubernetes load balancer to expose MATLAB Job Scheduler to MATLAB clients outside the cluster
+loadBalancerAnnotations: {} # Annotations to use for the load balancer
 
 # Parallel pool proxy settings
 poolProxyBasePort: 30000 # Base port for parallel pool proxies
@@ -94,6 +105,9 @@ matlabPVC: "" # Name of a PVC that contains a MATLAB Parallel Server installatio
 jobManagerUsesPVC: false # If true, the job manager container mounts the MATLAB Parallel Server installation from the PVC rather than using the jobManagerImage parameter
 additionalMatlabPVCs: [] # Names of PersistentVolumeClaims containing installations of older MATLAB Parallel Server releases, specified as an array
 
+# Worker pod settings
+additionalWorkerPVCs: {} # Additional PersistentVolumeClaims to mount on worker pods, specifed as a map of claim names to mount paths
+
 # Specify the maximum number of workers that the cluster can automatically resize to in your custom values.yaml file.
 # maxWorkers: 32