Get Started with Accelerator

In this section, it is assumed that Altair Accelerator has been installed and that Accelerator is ready to start. If Accelerator has not yet been installed, instructions are available in the Installation Guide.

The command to start and stop Accelerator is ncmgr.

  1. Before starting the commands, verify the following:
    1. SSH/RSH works to the slaves for taskers.tcl to work.
    2. Review the usage information for ncmgr
      
      vncmgr: Usage Message
        
                This program manages the vovserver for NetworkComputer.
                Copyright (c) 1998-2019, Altair Engineering.
        
        USAGE:
                ncmgr help|info|reset|start|stop|cm [OPTIONS]
        
        ACTIONS:
                info   [-queue|-q <name>] [-v]
                reset  [-soft | -hard | -h ]
                start  [-dir <server_working_dir>] [-force] [-queue|-q <name>]
                       [-port <port> ] [-webport <port>] [-roport <port>]
                       [-dbhost <host>] [-dbroot <path>] [-dbport <port>]
                       [-prod nc|wx] [-basequeue <name>]
                       The default <server_working_dir> is
                       /home/o/sandbox/jira/0776/trunk68775/201901/gcc921/vov/vnc.
                       This is the parent of the configuration (.swd) directory for
                       the queue.
                stop   [-force] [-freeze] [-queue|-q <name>] [-writeprdir <dirname>]
                       -force       Do not prompt for confirmation
                       -freeze      Instruct taskers to keep running and wait for a
                                    new server
                       -writeprdir  Writes the PR file to the specified directory
                       (which is created if necessary)
                cm     [-queue|-q name] <ACTION> [ARGUMENTS]
                       Configuration Management. Pass "help" for detailed usage.
        
        EXAMPLES:
                % ncmgr
                % ncmgr -h
                % ncmgr start -queue vnc2
                % ncmgr start -port 6699 -queue vnc99
                % ncmgr info
                % ncmgr reset -soft
                % ncmgr reset -hard
                % ncmgr cm help
        
        EXAMPLE TO STOP AND RESTART SERVER:
        
                % ncmgr stop -freeze
                % ncmgr start -force
                % ncmgr stop -freeze -force -writeprdir /tmp/abc123
        
        
        
    The default Accelerator queue is named vnc, which can be started with the command shown below.
    Note: The information displayed is acquired in a role account, not a user account. A role account provides access to the data that is stored in a central location that all tasker servers can see.
    % ncmgr start
    ncmgr: message: Checking the license...
    ncmgr: message: ... the license is good.
    ncmgr: message: Starting Accelerator
    ncmgr: message:   with name    vnc
    ncmgr: message:   on host      hostname
    ncmgr: message:   in directory /remote/release/vov/vnc
    ncmgr: message:   as user      username
    Do you want to proceed? (yes/[no]) > yes
    ncmgr: message: Updating config file '/remote/release/vov/...../vnc.tcl'
    ncmgr: message: Waiting for server to be ready ...
    ncmgr: message: Sanity check...
    ncmgr: message: Accelerator vnc@hostname is ready.
    The command below is equivalent to ncmgr start above, due to the default values of the working directory and the queue name.
    % ncmgr start -queue vnc -dir $VOVDIR/../../vnc
    ...output omitted...
  2. Optional: Create any additional queues. The name of each queue must begin with vnc.

    On any given LAN, it is recommended to use a single queue to coordinate all resources, CPUs and licenses. This will minimize the management overhead and maximize the resource utilization, and easier for the users.

  3. Once the server has been started, proceed with the browser interface to complete the setup. Connect to http://hostname:port/cgi/setup.cgi
    Where the hostname is the Accelerator server host name and the port number is one of:
    • the web port that was specified in the ncmgr command line
    • if no web port was specified, the single main port that was specified in the ncmgr command line
    • if no single main port was specified, the single main port that was chosen from one of the dynamic port modes, as shown at the end of the ncmgr start command output

    Alternatively, proceed with the manual configuration that follows.

  4. The ncmgr command creates a new directory with the default configuration files. Go to that directory: % cd $VOVDIR/../../vnc.
  5. Edit the taskers file vnc.swd/taskers.tcl and modify the host list to include the names of hosts in your network. This file is in Tcl syntax; insert the host names between the left and right braces of themainComputeServers and usedWhenIdle lists.

    The default taskers file differentiates between compute servers and workstations. The first type refers to machines that are available all the time. The second type applies to machines that should be available only if their console has been idle (no keyboard or mouse activity) for at least 10 minutes. Workstations accept only jobs that are expected to take less than 5 minutes.

  6. Put a few server and workstation names in the file. You will be able to refine the list and the configuration later.
  7. The security file vnc.swd/security.tcl determines who may connect to the system and from which host. By default, the owner of the vovserver has ADMIN privilege from all hosts. Check the file vnc.swd/security.tcl to verify the login name is correct and add any others needed.
  8. Check the file vnc.swd/security.tcl to verify that the following lines exist:
    # All users can connect from anywhere
    vtk_security + USER +
    
    # You want to be ADMIN
    vtk_security YOURNAME ADMIN +
    
    # Root needs to have ADMIN privileges, to permit vovslaveroot to connect
    vtk_security root ADMIN +
    
    # The cadmgrs VovUserGroup needs to have ADMIN privileges
    vtk_security -group cadmgrs ADMIN +
  9. If the taskers.tcl file was manually changed to add hosts, start any additional ones for Accelerator:
    % ncmgr reset -taskers

    Depending on their number, it may take a minute or more to start all taskers. The command starts them in groups to avoid loading the network too heavily.

  10. Check the Accelerator system status with the commands:
    % ncmgr info
    % nc hosts
    % nc cmd vsi

    The first two commands show information about the farm slave hosts. The last command shows information about the vovserver.