ResultsSet Cache Not Synchronized Between WCSites Cluster Nodes Due to Different Datasource Names (Doc ID 1984797.1)

Last updated on JUNE 20, 2023

Applies to:

Symptoms

In a 2-node cluster, update an asset in one node does not invalidate the ResultSet cache on the other node. (Other caches are working fine.)

A simple test case is to update the name/description of an asset on node1, and observe stale asset name/description on node 2 on the Inspect screen.

InCache/network communication is not problem here as DEBUG/TRACE messages shows incache notifications are able to reach to the other node.

For example, with the following logger enabled:

• net.sf.ehcache.distribution=TRACE
• net.sf.ehcache=TRACE
• com.fatwire.logging.cscache.notification=DEBUG

node1.log:

[2014-11-20 17:20:00,245 CET] [DEBUG] [WebContainer : 4] [net.sf.ehcache.Cache] Cache: jndi/datasourceNode1myAssetType store hit for jndi/datasourceNode1-SELECT ID FROM MYASSETTYPE WHERE ID=?[Number of positions=1] data [{0=1360680831344}]

node2.log:

[2014-11-20 17:21:08,118 CET] [DEBUG] [WebContainer : 0] [net.sf.ehcache.Cache] Cache: cached-catalogs store hit for jndi/datasourceNode2siteentry
...
[2014-11-20 17:20:41,244 CET] [DEBUG] [CP Connection(8)-{IP_Address}] [sf.ehcache.distribution.RMICachePeer] RMICachePeer for cache notifier: remote remove received for key: InvalidationNotification{cacheName='linked-cache', invalidationData=InvalidationData{deps=[jndi/datasourceNode1myAssetType], time=1416500441241}, key='18415852@MX03010016D0046-1296616724'}

Changes

When the cluster was set up, even though both nodes are pointing to the same database, they are using different datasource name.

For example:

node1: datasourceNode1

node2: datasourceNode2

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.