+

Search Tips   |   Advanced Search

File store high availability

A messaging engine in a cluster bus member can fail over between servers in the cluster, but the messaging engine must be able to access the saved state in its file store from any of the cluster servers.

File stores can be used in highly available environments. To make a file store highly available, use hardware or software facilities that can maximize the availability of the file store data, for example, a storage area network (SAN).

We must ensure that the directories containing the log file and store files can be accessed from all members of the cluster, using the same directory name.

We can test whether a shared file system can support the failover of messaging engines by running the File System Locking Protocol Test for WebSphere Application Server. To run the test see, IBM File System Locking Protocol Test for WebSphere Application Server.

WebSphere Application Server supports two styles of file system access to enable high availability:

Cluster-managed file system

This style of file system access uses high availability clustering and failover of shared disks to ensure that the directories for the file store are accessible from the server currently running the messaging engine. The directories are located on file systems in the shared disks, and high availability cluster scripts are used to mount the file systems on the node containing the server running the messaging engine.

Networked file system

This style of file system access allows remote files to be accessed over a network. The most popular protocols for accessing remote files are Common Internet File System (CIFS) and Network File System (NFS). Version 4 of NFS supports automated failover to ensure access locking. Access locking ensures the integrity of the log files; that is, only a single client process can access the log at a time.

It is important to check that the file system configuration is correct, because it cannot be checked by the WAS configuration system or messaging engine. Errors only surface at run time, so thorough failover testing is recommended.

Further information: The requirements for enabling access to the directories for the file store are similar to those for enabling access to the recovery log in a cluster. For more information see the following article: Transactional high availability and deployment considerations in WebSphere Application Server V6 .


Related concepts

  • Message stores
  • File stores