Description

When setting up File Server targets for archiving, the configuration allows the targets to be enabled for Pass-through recall, and a Cache location must be defined within the Properties of the File Server target.

Once the cache location is defined, the Placeholder Service start-up routine performs validation of the settings to enable the Cache location, as well as any Volumes that are configured for archiving and were enabled for Pass-through recall via the Vault Administration Console (VAC).

The same concept applies to File Server targets that are hosted on Windows Failover Clusters as File Server Roles.
The important point to consider is that where multiple File Server Roles may be active on one node, the FSA Agent will be responsible for processing recall requests from data stored on all shared disks that are mounted on the same node.
However, the FSA Agent cannot process Pass-through recall requests using different Cache locations configured on shared disks that are controlled by each File Server Role’s resource group.

The correct configuration is to define the Cache location for each clustered File Server target using a local drive on the nodes. As a general recommendation, the Cache location should not be hosted on the system drive, therefore an additional disk must be added to each participating node and the same Drive letter\Folder name must be created on all nodes. When a failover occurs, taking a File Server Role to another node, the Cache location will continue to work using the same folder on the other node.
 
The Setting up File System Archiving document states the following:
Link to Documentation - https://www.veritas.com/support/en_US/article.100040095

About configuring pass-through recall for a file server cluster
Note that if you configure pass-through recall for a file server cluster, all the cluster nodes must use identical pass-through recall settings.
In the file server properties for the target virtual file server, make sure that the pass-through recall settings are configured as follows:

• The "Configure pass-through recall" setting is checked.
• The disk cache location is a local path such as C:\FSACacheFolder. This path must be valid for a local disk on each cluster node.

Note:

If the cluster configuration supports only one active node, you may alternatively specify a location on the cluster's shared disk. For example, you can use a shared disk location for an A/P, A/P/P, or A/P/P/P configuration, but not for an A/A/P configuration, where A represents an active node and P represents a passive node.

• The disk cache size is specified. We recommend that you make the cache size as large as possible.

 
Currently, it is not recommended to use Pass-through recall method for File Server targets that need to service a high number of recall requests daily. This is due to the resulting high number of handles and corresponding memory consumption required to maintain the content in the Cache location.
The methods used to maintain the content in the Cache with the associated high number of handles is being enhanced and with the release of EV 12.2 a new approach will be implemented allowing for better performance and more efficient use of system resources.

Depending on the number of File Server Roles that may be active on one node at any time, the additional recommendation is to increase the size of the Cache location accordingly.
If the Cache location size is not sufficiently large to accommodate the normal amount of recall requests, this would trigger the Cache Clean-up routine more often, therefore leading to potential performance impact due to the unnecessary extra processing involved.

When planning the size of the Cache location for a cluster node, the minimal capacity should follow the rule:

  Disk Cache = 1GB X R
 
Note:
Where the combined Disk Cache size is less than the sum of Roles, the processing may result in higher disk bandwidth and CPU consumption.
Whilst not used, the Disk Cache on Passive cluster nodes is necessary for when a failover occurs.
Depending on different factors, such as volume sizes, number of placeholders and access patterns, further customisations may be required to determine the optimal size of the Cache location.
 

As an example:
On each node, add an additional local disk and mount with the same drive letter. Then create a folder on each node like in:
D:\ LocalPTCacheLocation

File Server Roles



VAC File Server Target settings

Recommendations for implementing File System Archiving (FSA) with Windows Failover Clustering File Server targets using Pass-through recall.