7.00.02 - Changing Aster Instance Partition Count - Aster Execution Engine

Aster Instance Installation Guide for Aster-on-Hadoop Only

Aster Execution Engine
July 2017

To change the number of partitions (vWorkers) in an Aster instance, perform these steps.

When performing this procedure, you may need to modify YARN memory allocation parameters or Aster configuration file settings, see “Determine the memory to be configured per vWorker” step under Set YARN Configuration Value Changes.
  1. As the root user, execute this command on queen node: ncli system softshutdown
  2. Verify that the Aster instance stopped successfully by issuing this command on the queen node: ncli node show

    The output from the ncli command should show only the queen and no workers.

  3. Change the partition count by issuing this command on the queen node: /home/beehive/bin/lib/configure/ConfigureNCluster.py --partition_count=new_partition

    new_partition is the number of desired partitions (vWorkers) in the Aster instance.

  4. In addition, update beehiveparams.cfg with the new partition count for the PARTITION_COUNT argument. However, the changed worker count does not need to be updated in /home/beehive/config/vWorkerDiskList.
  5. Start the Aster instance by issuing this command as the root user: ncli system softstartup
  6. Verify that the Aster instance restarted successfully by issuing these commands on the queen node: asteryarn --action=STATUS_INSTANCE yarn node --list --all

    The output from the asteryarn command should show all workers in Active state and the output from the yarn command should show one container for each node. Note that one of the nodes will have an additional container which is created for the Application master.