Caching issues on our production Portal
We have noticed an issue with our production Portal and I have not yet found the root cause.
After being up for a while, it seems that the users have cached information that is being retrieved. Sometimes, this is not what was requested.
We first noticed this when we had applied transports to the Portal. The Production Portal consists of a CI and one additional application server on another VM server. If we did not stop / restart the Portal after transports, the original objects would display as opposed the new, changed ones received with the transport.
Today, we ran into a different situation. We had not sent any changed objects / transports into the production Portal. Users were displaying a course description from the course catalog of the E-Learning solution. The system would correctly display the first course chosen. If they then wanted to view the course description of another course in the catalog, regardless of what was selected, the system displayed the course description for the first course that had been chosen. It seemed the first page was cached and the system then displayed the cached information on subsequent requests.
The users could log off and back into the Portal but received the same course description from that first chosen course. My users even tried using a different PC to access the course information and continued to see the original, first course description.
So far, I have been unable to find a way to reset the cached information at the user level. Since the same information appears if the user accesses through a second computer, I do not believe the info is cached at the PC level but rather at the user level within the Portal.
Has anyone seen this type of situation before? Is there a setting that tells the system to not use cached data for a selection? Is there a place to reset user cache without having to bounce the entire Portal?
And then has anyone seen this elsewhere and could offer suggestions as to the root cause of the problem? I would rather resolve why this happens in the first place as opposed to clearing cache on a schedule or bouncing the Portal on a regular basis.