Our search indexes in RAM seem to be getting corrupted.  When you do a search and get a list of results you will get an "Asset cannot be found" error when you click on some of the assets.

Typically, we schedule an outage, bring down the WAS servers, delete the index folders and then restart everything so that the indexes rebuild.  Starting and stopping WAS doesn't take long but rebuilding the indexes takes 4 hours or more.

We have a clustered environment with 3 WAS servers.

I was wondering if there is anything that can be done to minimize this corruption?  Is it a common problem?  What influences it?