These steps are required for cluster configuration for Share. If you are using an HTTP
load-balancing mechanism in front of a clustered installation, ‘sticky’ routing must be enabled
for the HTTP requests made by the Share tier to the repository tier (the
/SkyVault application).
This can be achieved in one of two ways:
-
Hard-wire each /share instance to its own
/SkyVault instance, bypassing the load balancer.
This can be achieved by populating each share-config-custom.xml file with a host name and port number that is not behind your load balancing mechanism.
-
If NTLM or Kerberos authentication is enabled with SSO, or external authentication is
enabled, then Share will use cookie-based sessions and you can configure your load
balancer to use sticky routing using the JSESSIONID cookie.
Depending on whether you want to enable NTLM or Kerberos authentication, or external authentication, perform either of the following two steps:
Note: If you are configuring a Share cluster within a load-balancing
environment, you must configure Hazelcast between the Share instances in order to provide
multicast messaging between the web-tier nodes. For more information, see Configuring Hazelcast between Share
instances.