Reg portal cache mechanism in the clustered envrironment

Santhoshkumar_Nagulanchi

 

Hi, 

We are experiencing a strange behavior after upgrading portal 7.4.3 to 8.2.2. Portal is not letting us add/delete a portlet to/from pannel or rearranging a portlet in pannel. We further noticed the issues is happening when both nodes are up. Its a clustered environment with 2 portal servers and We brought down one of the nodes and tested,we can rearrange the portlets,delete or add the portlet.When both nodes are up it stopped working. 

 

In the logs  found the following error.

[[ACTIVE] ExecuteThread: '0' for queue: 'weblogic.kernel.Default (self-tuning)'] ERROR com.vignette.portal.website.admin.internal.display - getDisplayState: unable to cache display state.

 

What this error means? How portal implemented/uses cache in the clustered envrironment? 

I would greatly appreciate your advice.

Al_Daniel

Hello Santhosh,

 

We identified a defect in the Portal 8.2 that can cause the behavior you are describing.  When upgrading a system that uses Oracle as the backend database, the installer will fail to convert lastModuleUpdate and lastLayoutUpdate columns in the PAGEMODULESETS table from DATE to TIMESTAMP datatype.  This can causes Portal to fall into a broadcaster loop because the DATE datatype does not record fractional seconds.  When that happens, your changes are undone by the duplicate broadcast messages before you have time to refresh the console.

Please check that these columns are using the correct datatype (TIMESTAMP) in your upgraded environment.

The fix for this installer issue will be found in the next major version of Portal.  In the meantime, the recommended solution is to manually change the datatype from DATE to TIMESTAMP.

 

Best Regards

Alex