To configure a distribution pool

To launch a Design-of-Experiment, Optimization, or Runner-Balance study from the command line on Linux, you need to tell the Simulation Job Manager (SJM), on the machine where the study is launched, about its distribution pool. You can do this using the sjmServer.py utility.

Note: Do not launch the Thinclient SJM from a location that is shared between nodes. In the event the SJM is shared between nodes, copy it to the local machine ( i.e. /tmp ) before launching it.

To ensure that the SJM is running on the Linux machine where you intend to launch the runstudy command (the primary execution node), type:
```
$ /sbin/pidof dssp_jobserver
```
If the SJM is running on the machine, this command returns the process identification descriptor.
Check that there is no firewall blocking the required TCP ports 43911 and 43921 and 42911.
Use the sjmServer.py utility to confirm that the correct application, e.g. Moldflow Insight 2019, is installed on all Windows and Linux machines to which child studies will be distributed. From the primary execution node type:
```
$ sjmServer.py query –s <hostname>
```
In addition, execute the above command on the localhost, to establish the size of the existing batch queue, the application name, and any configured distribution pool.
To inform the SJM on the primary execution node of its distribution pool, create a text file that contains the hostnames, or IP addresses, of each distributed node. Use one line per node. Specify the primary execution node using the word "LOCALHOST", not by its IP address or hostname. For example, a file called pool.txt shown below, specifies a distribution pool with two child nodes:
```
LOCALHOST
192.168.2.3
fred.computer.domain.com
```
To set the distribution pool of the SJM, type:
```
$ sjmServer.py set –a <Application name> -d pool.txt
```
Once executed, this displays the server status and lists the hostnames specified in your distribution pool text file.
Use the runstudy command to launch an analysis.
To adjust the batch queue limit (how many jobs can be run in parallel), type:
```
$ sjmServer.py set –a <Application name> -n 5
```
for each machine, and change the integer from 5, as appropriate.