Preview generation is usually a CPU demanding task that might produce server slowness in some cases. If you have enough resources available you will be interested in changing the preview generation speed to take advantage of the available resources.
- Check the database options configured for the volume(s) containing PDF files, there is a key parameter that affects the PDF preview generation called Page limit for multi-page PDF Previews for more details you can check this article. Reconsider changing the value for the mentioned setting if the previews reducing it to the minimum value required by the business needs. If your organization does not store PDF files in Xinet, you can skip this step.
- Open the configuration file located in the following path
- Identify the specific type of asset preview generation that you would like to change, you have the following options
Asset type Command QuarkXPress files quarkview PDF files pdfsync Video files movsync Office files officesync
- [16-31] has to be replaced with a unique number between 16 and 31
- timeout has to be replaced with the number of minutes before the preview generation of an asset timeouts and is marked as failed (no preview for that asset will be generated)
- multi has to be replaced with the number of previews being generated in parallel
- command has to be replaced with the specific command for the preview generation, refer to the table above
- In this example, we will customize the PDF preview configuration, the default configuration looks like the following
18 $v/bin/pdfsync -fid$i -volflags$f
Supposing that you would like to set a timeout of 100 minutes before the preview for a file is canceled and also 5 parallel PDF previews being generated, the configuration will end looking like the following
18(100,5) $v/bin/pdfsync -fid$i -volflags$f
- Save the changes applied to the file, the preview generation process will reload the configuration automatically
- Monitor the CPU usage and repeat the actions on this article in case you still do not have the desired progress and if you still have available resources
Note: there is no a recipe or a particular recommendation to achieve a perfect balance between the timeout and the parallel previews being generated, we suggest to increase the parameters in a controlled way and monitor the server resources