Reduce am resource (#2072)

* reduce am resource to 1G * fix test * increase yarn virtual memory limit * move hard-code to config * add other am resource config
microsoft · Jan 31, 2019 · 1e6c84d · 1e6c84d
1 parent 50c26cb
commit 1e6c84d
Show file tree

Hide file tree

Showing 3 changed files with 40 additions and 0 deletions.
diff --git a/src/hadoop-node-manager/deploy/hadoop-node-manager-configuration/yarn-site.xml.template b/src/hadoop-node-manager/deploy/hadoop-node-manager-configuration/yarn-site.xml.template
@@ -27,6 +27,16 @@
         <description>Number of memory that can be allocated for containers.</description>
     </property>
 
+    <property>
+        <name>yarn.nodemanager.vmem-pmem-ratio</name>
+        <value>6</value>
+        <description>
+            Ratio between virtual memory to physical memory when setting memory limits for containers.
+            Container allocations are expressed in terms of physical memory, and virtual memory usage
+            is allowed to exceed this allocation by this ratio.
+        </description>
+    </property>
+
     <property>
         <name>yarn.resourcemanager.max-completed-applications</name>
         <value>1000</value>

diff --git a/src/rest-server/src/config/launcher.js b/src/rest-server/src/config/launcher.js
@@ -43,6 +43,12 @@ let launcherConfig = {
   jobDirCleanUpIntervalSecond: 7200,
   jobConfigFileName: 'JobConfig.json',
   frameworkDescriptionFilename: 'FrameworkDescription.json',
+  amResource: {
+    cpuNumber: 1,
+    memoryMB: 1024,
+    diskType: 0,
+    diskMB: 0,
+  },
 };
 
 launcherConfig.healthCheckPath = () => {
@@ -125,6 +131,24 @@ const launcherConfigSchema = Joi.object().keys({
     .default('JobConfig.json'),
   frameworkDescriptionFilename: Joi.string()
     .default('FrameworkDescription.json'),
+  amResource: Joi.object().keys({
+    cpuNumber: Joi.number()
+      .integer()
+      .min(1)
+      .default(1),
+    memoryMB: Joi.number()
+      .integer()
+      .min(1024)
+      .default(4096),
+    diskType: Joi.number()
+      .integer()
+      .default(0),
+    diskMB: Joi.number()
+      .integer()
+      .min(0)
+      .default(0),
+  }),
+
 }).required();
 
 const {error, value} = Joi.validate(launcherConfig, launcherConfigSchema);

diff --git a/src/rest-server/src/models/job.js b/src/rest-server/src/models/job.js
@@ -466,6 +466,12 @@ class Job {
         'queue': virtualCluster,
         'taskNodeGpuType': gpuType,
         'gangAllocation': true,
+        'amResource': {
+          'cpuNumber': launcherConfig.amResource.cpuNumber,
+          'memoryMB': launcherConfig.amResource.memoryMB,
+          'diskType': launcherConfig.amResource.diskType,
+          'diskMB': launcherConfig.amResource.diskMB,
+        },
       },
     };
     for (let i = 0; i < data.taskRoles.length; i++) {