PrizmDoc Server uses an internal component called Office Conversion Service for converting Office documents, such as Word, PowerPoint and Excel.
The settings listed here affect LibreOffice only.
To customize Office Conversion Service settings, edit the central configuration file. To achieve optimal performance for document conversion, please consider adjusting values for the following parameters:
| Example |
Copy Code
|
resourceUsage.ocs.numInstances: 4
resourceUsage.ocs.numThreads: 8
resourceUsage.ocs.numPorts: 16
|
- resourceUsage.ocs.numInstances - Represents the number of office document converter processes the service can run. It is recommended to use as many Office conversion instances on the system as there are CPU cores available.
- resourceUsage.ocs.numThreads - Number of working threads created by service to handle document processing requests. Recommended value: 2 * resourceUsage.ocs.numInstances. Setting to 1 or values much smaller than resourceUsage.ocs.numInstances, may make the service less responsive. Setting to values greater than 2 * resourceUsage.ocs.numInstances will not improve performance but will increase resource usage.
- resourceUsage.ocs.numPorts - Represents the number of ports which can be used internally for communication with the Office Conversion instances. Recommended value: 4 * resourceUsage.ocs.numInstances.
Automatic Settings
The Office Conversion Service can be configured to automatically use optimized values for above mentioned parameters. Just set the value "auto" to the required parameter:
| Example |
Copy Code
|
resourceUsage.ocs.numInstances: auto
resourceUsage.ocs.numThreads: auto
resourceUsage.ocs.numPorts: auto
|
The logic for handling "auto" parameters:
- Read resourceUsage.ocs.numInstances. If it is "auto", then assign to it the number of virtual CPU cores multiplied by 2 and divided by 3. Otherwise use explicit integer value. For example:
- On a 4 core machine with hyper-threading that reported 8 virtual cores, 2/3 of 8 is 5.3333, which rounds to 5. So 5 Office conversion instances would be used.
- On a 4 core machine without hyper-threading that reported 4 virtual cores, 2/3 of 4 is 2.6666, which rounds to 3. So 3 Office conversion instances would be used.
- Then read resourceUsage.ocs.numThreads. If it is "auto", then assign to it resourceUsage.ocs.numInstances * 2. Otherwise use explicit integer value.
- Then read resourceUsage.ocs.numThreads. If it is "auto", then assign to it resourceUsage.ocs.numInstances * 4. Otherwise use explicit integer value.
The Office Conversion Service performs case sensitive comparison for "auto" value.
The following configuration properties have been deprecated and will be removed in a future release. Alter these properties only if not using the central configuration file.
To customize Office Conversion Service settings, edit the Watchdog configuration file located in the following default directories:
- Windows: C:\Prizm\PCCIS\Watchdog\watchdog.config
- Linux: /usr/share/prizm/pccis/Watchdog/watchdog.config
To achieve optimal performance for document conversion, please consider adjusting values for the following parameters:
| Example |
Copy Code
|
"officeConversionService": {
...
"officeInstanceCount":4,
"threadCount":8,
"officePortCount":16,
...
},
|
- officeInstanceCount - Represents the number of office document converter processes the service can run. It is recommended to use as many Office conversion instances on the system as there are CPU cores available.
- threadCount - Number of working threads created by service to handle document processing requests. Recommended value: 2 * officeInstanceCount. Setting to 1 or values much smaller than officeInstanceCount, may make the service less responsive. Setting to values greater than 2 * officeInstanceCount will not improve performance but will increase resource usage.
- officePortCount - Represents the number of ports which can be used internally for communication with the Office Conversion instances. Recommended value: 4 * officeInstanceCount.
Automatic Settings
The Office Conversion Service can be configured to automatically use optimized values for above mentioned parameters. Just set the value "auto" to the required parameter:
| Example |
Copy Code
|
"officeConversionService": {
...
"officeInstanceCount":"auto",
"threadCount":"auto",
"officePortCount":"auto",
...
},
|
The logic for handling "auto" parameters:
- Read officeInstanceCount. If it is "auto", then assign to it the number of virtual CPU cores multiplied by 2 and divided by 3. Otherwise use explicit integer value. For example:
- On a 4 core machine with hyper-threading that reported 8 virtual cores, 2/3 of 8 is 5.3333, which rounds to 5. So 5 Office conversion instances would be used.
- On a 4 core machine without hyper-threading that reported 4 virtual cores, 2/3 of 4 is 2.6666, which rounds to 3. So 3 Office conversion instances would be used.
- Then read threadCount. If it is "auto", then assign to it officeInstanceCount * 2. Otherwise use explicit integer value.
- Then read officePortCount. If it is "auto", then assign to it officeInstanceCount * 4. Otherwise use explicit integer value.
The Office Conversion Service performs case sensitive comparison for "auto" value.
