The METACLEANSE procedure removes orphaned Property and PropertyType objects from all Foundation and custom repositories. This procedure is designed to be used only against these two objects and should not be used against any other object type. If orphaned objects exist, the size of the Foundation or custom repositories will be reduced once PROC METACLEANSE is run.

Before you follow the steps below, you might want to use the CONTAINERSTATS routine to generate detailed statistics on the size of your metadata before you begin. The CONTAINERSTATS output provides detailed statistics on the overall size of metadata along with a breakdown of how the metadata is distributed across all containers (data sets). Information on the CONTAINERSTATS routine can be found at the top of page 111 in Chapter 14 "Managing Memory for Performance" of the SAS^® 9.1.3 Intelligence Platform System Administration Guide.

To execute the metadata cleanup processes that find and remove metadata content that is no longer in use (orphaned) follow these steps:

1. Back up all metadata. Ensure that a valid metadata backup (%OMABAKUP) has been run and saved to a secure location.

   CAUTION: SAS recommends that you use %OMABAKUP to perform your metadata backups. If you choose to use system utilities to perform your backup instead, be aware that you *must* stop or properly pause the metadata server before running the backup utility. This ensures that any changes made to metadata just prior to the back have time to be committed to disk, and it releases locks that the metadata server has on the disk data sets, allowing the host utility access to those data sets.

2. Use PROC METACLEANSE to remove orphaned Property and PropertyType objects from all Foundation and custom repositories. Save the following code to a file, then modify it so the options (SERVER=, PORT=, USERID=, and PASSWORD=) reflect the site–specific values in both PROC METACLEANSE invocations. This must be run using the unrestricted user account:
   PROC METACLEANSE SERVER="your-metaserver" PORT=8561 USERID="sasadm" PASSWORD="sasadmpw" REPOSITORY="Foundation" PROTOCOL=BRIDGE TYPE="Property" /* Do not change */ VERBOSE CLEANUP; RUN; PROC METACLEANSE SERVER="your-metaserver" PORT=8561 USERID="sasadm" PASSWORD="sasadmpw" REPOSITORY="Foundation" PROTOCOL=BRIDGE TYPE="PropertyType" /* Do not change */ VERBOSE CLEANUP; RUN;

   Execute the modified code (in a SAS Session, SAS^® Enterprise Guide^® Code window, or a SAS^® Data Integration Studio Source Editor window) against the Foundation repository.

   If you have any custom repositories defined in metadata, modify the REPOSITORY= option and enter the custom repository name, then rerun the modified code. Repeat this process until all custom repositories have been processed.

   WARNING: Do not run this process against project repositories. Only run this against Foundation and custom repositories.

   CAUTION: SAS Technical Support recommends that you process only Property and PropertyType object types with METACLEANSE and the CLEANUP option. Do not run this process with any other values for TYPE=. The METACLEANSE procedure is not aware of all usage scenarios employed by various products when it comes to many metadata objects; therefore, it may report objects as orphaned when they are, in fact, valid and in use. If you delete objects that fit into this category you might cause irreparable damage to the metadata; thus, requiring a restore from backup.

3. Remove orphaned WorkTable objects from Foundation and custom repositories. If your site uses the SAS Data Integration Studio product there is a cleanup tool that can be installed and launched from within the product to find and remove orphaned WorkTable objects in metadata. The number of WorkTable objects (and associated subordinate metadata) can be substantial if you delete SAS Data Integration Studio jobs. Perform the following to find and remove these objects.
   1. Download the sas.dbuilder.metadataclean2.jar: ftp://ftp.sas.com/techsup/download/blind/sas.dbuilder.metadataclean2.jar
   2. Save the JAR file to the following location on a SAS Data Integration Studio client machine:

      C:\Program Files\SAS\SASETLStudio\9.1\plugins

      This assumes you have the product installed in the default location. If that is not the case, save the JAR file to the actual installation location's plugins folder.

   3. Launch SAS Data Integration Studio and log into the Foundation repository as the unrestricted user.
   4. Select Tools ► Cleanup Job Metadata from the product menu. Select Yes to the question regarding metadata backup (assuming you did perform the backup as outlined above!). The utility will then perform an analysis on the repository and determine how many orphaned objects can be deleted (this may take a while to run, so just be patient). Next, it will display a window with information on the number of objects that it has determined can be deleted. Select Yes, and those objects will be removed.
   5. If custom repositories exist on the server, repeat the process outlined above by logging into each of the custom metadata repositories as the unrestricted user and following the same steps.

4. Perform another metadata backup (%OMABAKUP) with the REORG=YES option. If you execute either of the cleanup processes listed above, you will want to perform another metadata backup with REORG=YES to remove deleted content from the metadata containers on disk. Then you can run the CONTAINERSTATS routine again to get updated metadata statistics and compare these results to those generated before cleanup to determine the new metadata size.
5. Use the example at the top of page 112 of the SAS^® 9.1.3 Intelligence Platform System Administration Guide to estimate the total amount of memory required for loading all metadata. This would approximate the maximum amount of memory the server would use if you were to access all metadata types while the server is running.

Operating System and Release Information