|
Points to note |
|
|
Points to note |
If the OrcaFlex models being processed by Distributed OrcaFlex use either Python external functions or post calculation actions then each Distributed OrcaFlex client machine needs the required version of Python installed. Similarly for post calculation actions using command scripts then programs used by the script need to be present on each client machine.
| Note: | MATLAB can’t be started by a Distributed OrcaFlex client service since it tries to open a console on the desktop – and this is not possible for an application running as a service. |
If the model file has the Skip save
setting enabled then the simulation file for that model will not be saved after the post calculation action has completed, also any auto save setting will be ignored.
When running on computers with a large processor count then Python processing will be slow as the Python interpreter is effectively single threaded. There is one Python interpreter per DOF client process, so it will be beneficial in this case to configure the client machine to start more DOF clients. See the DOFMaxThreadCountPerProcess registry setting for the DOF client.
Installing an upgrade to OrcaFlex on machines that also have Distributed OrcaFlex installed does not update the default OrcFxAPI.dll (used when processing jobs). The location of the default OrcFxAPI.dll is defined on each of the client machines by the registry setting DOFDefaultDLLFileName which is a string containing the full UNC path to the DLL.
To update the default DLL used by Distributed OrcaFlex, you need to manually copy the new version of the DLL to the location specified by the DOFDefaultDLLFileName setting.
If the location of the default OrcFxAPI.dll needs to be changed, then it is very important that all the client machines running the DOF viewer have their registry setting updated to reflect the new location. This setting can be seen in the DOF viewer in the settings page when submitting jobs.