CheckArticles scheduled job is clearing the portal cache when sharding is enabled

• Type: Bug
• Status: Closed
• Resolution: Fixed
• Affects Version/s: 6.2.10 EE GA1, 6.2.X EE
• Fix Version/s: 6.2.X EE
• Component/s: Core Infrastructure, Core Infrastructure > Sharding
• Labels:
  • 62-ee-sp19
  • ee-6.2.x-only
  • liferay-fixpack-portal-140-6210
• Branch Version/s:
  6.2.x
• Backported to Branch:
  Committed
• Story Points: 1.5
• Fix Priority: 4
• Git Pull Request: https://github.com/brianchandotcom/liferay-portal-ee/pull/16192

This job is clearing the portal cache each 15 minutes (this is the default value configured via property journal.article.check.interval=15), which has a lot of impact in the portal performance.

Note: Only reproducible with a DB Sharding enabled

Steps to check the issue

1. Configure a db sharding environment folloking link Database Sharding
2. Add next property to portal-ext.properties for doing easier the test: journal.article.check.interval=2
3. Restart the server and create a new portal instance in the second shard from Control Panel > Instances
4. Got to Control Panel > Server Admin. > Log levels > add category: com.liferay.portlet.journal.service.impl.JournalArticleLocalServiceImpl ALL
5. Run jconsole in a terminal, if your server is running in localhost connect to the local process org.apache.catalina.startup.Bootstrap
6. In the portal - Run the groovy script:

for (def g: com.liferay.portal.service.GroupLocalServiceUtil.getGroups(-1, -1)) {
out.println (com.liferay.portal.service.GroupLocalServiceUtil.getGroup(g.getGroupId()).getName())
}

7. In JConsole Go to the MBeans and navigate to net.sf.ehcache - CacheStatistics - liferay-multi-vm-clustered - com.liferay.portal.kernel.dao.orm.EntityCache.com.liferay.portal.model.impl.GroupImpl
8. Get the Attribute - ObjectCount - Refresh, you will check the number of cached layouts, for example: 10
9. Check the logs until you view a trace like this (it appears each 2 minutes and appears repeated by each shard):

14:53:25,689 DEBUG [liferay/scheduler_dispatch-4][JournalArticleLocalServiceImpl:5801] Expiring 0 articles

10. In JConsole, refresh the same attribute ObjectCount after the trace appears.

Results The cache has been cleared and the number of objects is 0 or 1 (may be there is any object but the most of the objects in cache has been cleared).
Expected Number of objects are the same.