Capping the number of persistent grants helps limit the data stored for persistent grants, especially in scenarios where clients frequently request authorization in a single context.

When PingFederate needs to record a new grant, it checks whether such creation will push the number of grants beyond the limit. If it does, PingFederate creates the grant and then removes just enough grants so that the number of grants is capped at the limit. This cleanup task starts from the oldest grant, expired or not, and continues forward if it needs to remove multiple grants. For performance reasons, this cleanup task also limits the number of grants it can remove per attempt. If it cannot remove all grants in excess of the limit, it removes what it can and repeats the process when PingFederate needs to record a new grant.

This cleanup runs on every engine node in a clustered PingFederate environment. Also, it does not replace the cleanup task or the PingDirectory plugin engineered to manage expired grants. Working together, they keep the size of the grant datastore under control.

The default limit is 100 grants per user, client, grant type, and authentication context. Depending on the storage platform, the default maximum number of grants that this cleanup task can remove per attempt varies.

This cleanup task is enabled on new installations. When upgrading from version 9.1 or an earlier version, it is disabled. You can enable it by editing an XML configuration file.

  1. Edit the configuration file relevant to your storage platform.

    This configuration file is located in the <pf_install>/pingfederate/server/default/data/config-store directory, as described in the following table.

    Storage platform Configuration file

    Database server

    org.sourceid.oauth20.token.AccessGrantManagerJdbcImpl.xml

    PingDirectory

    org.sourceid.oauth20.token.AccessGrantManagerLDAPPingDirectoryImpl.xml

    Microsoft Active Directory

    org.sourceid.oauth20.token.AccessGrantManagerLDAPADImpl.xml

    Oracle Unified Directory

    org.sourceid.oauth20.token.AccessGrantManagerLDAPOracleImpl.xml
  2. Locate for the following comments. 
    ...
<!--
    Maximum number of persistent grants allowed to store in the database per user,
    client and grant type and authentication context qualifier.

    Setting this to a value <= 0 will turn this limit off
    Default configuration:
    <c:item name="maxPersistentGrants">100</c:item>
-->
<c:item name="maxPersistentGrants">100</c:item>
<!--
    Maximum number of persistent grants to delete when max allowed is reached
    during new grant creation.

    Setting this to a value <= 0 will turn this limit off
    Default configuration:
    <c:item name="maxPersistentGrantsToRemoveBatchSize">n</c:item>
-->
<c:item name="maxPersistentGrantsToRemoveBatchSize">n</c:item>
...

    The maxPersistentGrants value represents the maximum number of grants based on a combination of user, client, grant type, and authentication context.

    The maxPersistentGrantsToRemoveBatchSize value represents the maximum number of grants that the cleanup task would remove per attempt. Its default value (n) varies depending on the storage platform, 50 for a database server and 10 for a directory server.

    Note:

    The maxPersistentGrants and maxPersistentGrantsToRemoveBatchSize items exist only on new installations starting with version 9.2. When upgrading from version 9.1 or an earlier version, the upgrade tools only insert the comments for reference.

  3. Optional: Adjust the maxPersistentGrants and maxPersistentGrantsToRemoveBatchSize values.

    Use integers only.

  4. To enable this cleanup task after upgrading from version 9.1 or an earlier version, insert the maxPersistentGrants and maxPersistentGrantsToRemoveBatchSize items into the configuration file.

    You can use the default values based on the inline comment. You can also adjust the values to suit the needs of your organization.

  5. Save your changes.
  6. Restart PingFederate.

    For a clustered PingFederate environment, perform these steps on the console node, and then click Replicate Configuration on System > Server > Cluster Management. You do not have to restart PingFederate on any running engine node.