Each metadata server in a cluster, including the master metadata server, is assigned a workload; a workload is the metadata and locks that a metadata server is responsible for providing metadata to clients whenever they request access. A fileset is the unit of workload for metadata servers. Data resides in filesets assigned to metadata servers.
During client setup, a client is given the address of one of the metadata servers for initial contact and metadata server cluster discovery. When the client issues a request to access data, it is automatically directed to the appropriate metadata server to obtain the metadata and locks required to access the data.
SAN File System can automatically balance the workload for you, by dynamically assigning filesets across all metadata servers.
You can also choose to control the workload manually by statically assigning the filesets to specific metadata servers. When assigning static filesets, consider the metadata-activity level of each fileset, and balance the filesets evenly across all metadata servers in the cluster. To ensure that each metadata server handles a share of the entire workload, assign at least one fileset to each metadata server. Creating additional filesets provides you with greater flexibility in assigning and reassigning filesets to achieve optimal results. You can statically assign a fileset to a metadata server or unassign a fileset at any time.
A spare metadata server is an idle metadata server that has no statically assigned filesets. You can reserve a spare metadata server for failover to take on the workload of another metadata server that goes offline and preserve the workload balance.
Parent topic: Cluster