Help Center/ MapReduce Service/ User Guide/ MRS Manager Operation Guide (Applicable to 2.x and Earlier Versions)/ Static Service Pool Management/ Configuring a Static Service Pool
Updated on 2024-04-11 GMT+08:00
View PDF

Configuring a Static Service Pool

Scenario

If you need to control the node resources that can be used by the cluster service or the CPU usage of the node used by the cluster in different time periods, you can adjust the resource base on MRS Manager and customize the resource configuration groups.

Prerequisites

  • After the static service pool is configured, the HDFS and YARN services need to be restarted. During the restart, the services are unavailable.
  • After a static service pool is configured, the maximum number of resources used by each service and role instance cannot exceed the upper limit.

Procedure

  1. Modify the system resource adjustment base.

    1. On MRS Manager, click System. In the Resource area, click Configure Static Service Pool.
    2. Click Configuration. The service pool configuration group management page is displayed.
    3. In the System Resource Adjustment Base area, change the values of CPU(%) and Memory(%).

      Modifying System Resource Adjustment Base limits the maximum physical CPU and memory resource percentage of nodes that can be used by the Flume, HBase, HDFS, Impala and YARN services. If multiple services are deployed on the same node, the maximum physical resource usage of all services cannot exceed the adjusted CPU or memory usage.

    4. Click Next.

      If you need to modify the parameters again, click Previous in the lower part of the page.

  2. Modify the default configuration group of the service pool.

    1. In the Service Pool Configuration table, set CPU LIMIT(%), CPU SHARE(%), I/O(%), and Memory(%) for the Flume, HBase, HDFS, Impala, and YARN services.
    • The sum of CPU LIMIT(%) used by all services can exceed 100%.
    • The sum of CPU SHARE(%) and I/O(%) used by all services must be 100%. For example, if CPU resources are allocated to the HDFS and Yarn services, the total CPU resources allocated to the two services are 100%.
    • The sum of Memory(%) used by all services can be greater than, smaller than, or equal to 100%.
    • Memory(%) cannot take effect dynamically and can only be modified in the default configuration group.
    1. Click in the blank area of the page to complete the editing. MRS Manager generates the correct values of service pool parameters in the Detailed Configuration area based on the cluster hardware resources and allocation information.
    2. You can click the edit icon on the right of Detailed Configuration to modify the parameter values of the service pool based on service requirements.

      In the Service Pool Configuration area, click the specified service name. The Detailed Configuration area displays only the parameters of the service. Manual changing of parameter values does not refresh the service resource usage. In added configuration groups, the configuration group numbers of the parameters that take effect dynamically will be displayed. For example, HBase: RegionServer: dynamic-config1.RES_CPUSET_PERCENTAGE. The parameter functions do not change.

      Table 1 Parameters of the static service pool

      Parameter

      Description
      • RES_CPUSET_PERCENTAGE
      • dynamic-configX.RES_CPUSET_PERCENTAGE

      Configures the service CPU percentage.
      • RES_CPU_SHARE
      • dynamic-configX.RES_CPU_SHARE

      Configures the service CPU share.
      • RES_BLKIO_WEIGHT
      • dynamic-configX.RES_BLKIO_WEIGHT

      Configures service I/O usage.

      HBASE_HEAPSIZE

      Configures the maximum JVM memory for RegionServer.

      HADOOP_HEAPSIZE

      Configures the maximum JVM memory of a DataNode.

      yarn.nodemanager.resource.memory-mb

      Configures the memory that can be used by NodeManager on the current node.

      dfs.datanode.max.locked.memory

      Configures the maximum memory that can be used by a DataNode as the HDFS cache.

      FLUME_HEAPSIZE

      Configures the maximum JVM memory that can be used by each Flume instance.

      IMPALAD_MEM_LIMIT

      Configures the maximum memory that can be used by an Impalad instance.

  3. Add a customized resource configuration group.

    1. Determine whether to automatically adjust resource configurations based on the time.

      If yes, go to 3.b.

      If no, go to 4.

    2. Click to add a resource configuration group. In the Scheduling Time area, click the edit icon. The time policy configuration page is displayed.

      Modify the following parameters based on service requirements and click OK.

      • Repeat: If selected, the resource configuration group runs repeatedly based on the scheduling period. If not selected, set the date and time when the configuration of the group of resources can be applied.
      • Repeat Policy: can be set to Daily, Weekly, and Monthly. This parameter is valid only when Repeat is selected.
      • Between: indicates the time period between the start time and end time when the resource configuration is applied. Set a unique time range. If the time range overlaps with that of an existing group of resource configuration, the time range cannot be saved. This parameter is valid only when Repeat is selected.
      • The default group of resource configuration takes effect in all undefined time segments.
      • The newly added resource group is a parameter set that takes effect dynamically in a specified time range.
      • The newly added resource group can be deleted. A maximum of four resource configuration groups that take effect dynamically can be added.
      • Select a repetition policy. If the end time is earlier than the start time, the next day is labeled by default. For example, if a validity period ranges from 22:00 to 06:00, the customized resource configuration takes effect from 22:00 on the current day to 06:00 on the next day.
      • If the repeat policy types of multiple configuration groups are different, the time ranges can overlap. The policy types are listed as follows by priority from low to high: daily, weekly, and monthly. The following is an example. There are two resource configuration groups using the monthly and daily policies, respectively. Their application time ranges in a day overlap as follows: [04:00 to 07:00] and [06:00 to 08:00]. In this case, the configuration of the group that uses the monthly policy prevails.
      • If the repeat policy types of multiple resource configuration groups are the same, the time ranges of different dates can overlap. For example, if there are two weekly scheduling groups, you can set the same time range on different day for them, such as to 04:00 to 07:00, on Monday and Wednesday, respectively.
    3. On the Service Pool Configuration page, modify the resource configuration of each service. Click the blank area on the page to complete the editing, and go to 4.

      You can click the edit icon on the right of Service Pool Configuration to modify the parameters. Click the edit icon in the Detailed Configuration area to manually update the parameter values generated by the system based on service requirements.

  4. Saves the settings.

    Click Save. In the Save Configuration dialog box, select Restart the affected services or instances. Click OK to save the settings and restart related services.

    Operation succeeded is displayed. Click Finish. The service is started.