IBM FileNet P8, Version 5.2.1            

Indexing is delayed because of low disk space

Text extraction consumes too much disk space during the Content Search Services (CSS) indexing process.

Symptoms

Depending on the amount of free disk space remaining, one of two possible warnings is written to the Content Platform Engine error log, p8_server_error.log. The following warnings are examples:

Causes

Content Platform Engine uses a temporary folder for extracting the text that will be indexed. By default, the temporary folder is configured to have a required minimum of at least 3 GB of free space for text extraction. (The default can be changed with a Java Virtual Machine (JVM) argument.)

For the first warning, the available disk space has fallen below the configured minimum of free disk space plus 2 GB (for example, below 5 GB when the default is 3 GB.) However, index requests are still processed.

For the second warning, the available disk space has dropped to the point that it is less than the configured minimum (3 GB by default). When this happens, dispatching of index requests is put on hold.

Resolving the problem

For a typical production environment, the temporary folder used for extracting text should be on a local file system with at least 5 GB of available disk space. If possible, free up disk space by removing files not related to Content Platform Engine.

If the default value for minimum disk space for text extraction has been overridden with the JVM argument com.filenet.cbr.minimumDiskSpaceBytes, verify that the specified value is not too high for the indexing workload. If it is, reduce the setting to a more appropriate value (but do not set this JVM argument to less than the default minimum of 3 GB).

If you cannot free up enough disk space, put the temporary folder in a file system with more available disk space. The performance of CSS can be improved if the folder is located on a RAM disk or other fast storage, such as a solid-state drive.



Last updated: October 2015
p8pcs029.htm

© Copyright IBM Corporation 2015.