Friday, October 25, 2013

db.FTSearch != Ubuntu Server

We ran into a db.FTSearch problem with a customer a few years ago.  Full indexed search failed on PDF and DOCX file attachments. For other attachments it worked fine. We were using Ubuntu 10.4 LTS.  Searching through we could only find a few references to db.FTsearch and Linux regarding these two file formats, but all the postings provided no resolution to the issue.  Fortunately, when we moved the database to production on a Windows server, the db.FTsearch for PDFs and DOCX files worked.  At that time, I thought it might have been our Linux server setup done incorrectly.

We recently had the same problem again using a different and newer installation of Ubuntu, 12.04 LTS server.  Further research and many attempts in changing configuration file did not help.  So we switch the server to CentOS as suggested by Bill Malchisky. Using the installation instructions and videos on Youtube from Devin Olsen, we installed CentOS 6.4 and now db.FTSearch works with PDFs and DOCX files. It seems that the FTsearch have issues with the Debian distribution of Linux.  Whether it is implementation of the KeyView filters not working or there is a bug I do not know.  This is beyond my Linux skills to track down.

Though I like Ubuntu alot, we will be switching all my servers to CentOS.  Lesson learned, if you need to perform db.FTSearch on databases that includes PDF or DOCX file attachments, do not use Ubuntu for the client or the server.