Distributed Search

MailArchiva Enterprise Edition v1.6 or greater supports distributed searching of indexes across multiple machines.

The primary objective of this feature is to enable efficient and timely search across extremely large archives. With MailArchiva, you can implement a search cluster.

Here is how to implement distributed search:

(1) Copy an old index to a second machine

(2) Mount an NFS share, accessible to both machines, on the volume store directory

(3) Install MailArchiva on the second machine

(4) Copy the MailArchiva server.conf file from the first machine

(5) Create a volume on the second machine

a) set archive store to be the mount point that refers to the shared NFS drive

b) set index to be the local index that was copied from the first machine

(6) On the second machine, click the publish volume for distributed search in the Volumes screen

(7) On the first machine, in the volume's index location, type "rmi//ipaddress" (where ipaddress is the ip address of the second machine) You will see that the volume status changes to "REMOTE". This means that the volume is pointing to the index on the external machine.

When you do a search on the first machine, it will use the processing resources of the second machine. The first search operation is slow because it has to initialize the connection. Once this is established, the process works very smoothly.

Note: You can setup as many machine as you like! The above is an example of two machines in operations

-- TWikiGuest - 24 Oct 2007