Re: Index with duplicate values

David, We experimented a similar problem with a customer. It was on Solaris. When our OPTCOMPIND parameter was set to the value of 2, then we could not even select the rows from the database, since and index existed on a certain column. We tought our index was corrupted, once we dropped the index , the engine actually behave normal (in that it returned the correct results, otherwise it always returned a different number of rows , say for example in the table there were 2,000 rows, the engine reported only 150 rows). I would first recommend you to use the key level locking , since it can avoid some locking problems (Asumming your index is being used, thus informix will actually lock the key of the index). Second, try to set your OPTCOMPIND value to 0 (if it is not set to that value, already), it is recommended for OLTP applications, such as yours. Hope this helps, Mario Estrada --------------------------------------Reply Separator------------------------------ -----Original Message----- From: David K. Killough <killougd@ix.netcom.com> To: informix-list@iiug.org <informix-list@iiug.org> Date: S'bado 19 de Septiembre de 1998 02:11 PM Subject: Re: Index with duplicate values > >Mario Estrada wrote in message <6tveml$ao$1@news.xmission.com>... >> >>First of all, >> >>It is recommended not to have an index on a column with few different >>values. > >OK. I'll buy that. But what do we do when we quick access based on the >value in that column. This is a tracking application, and the column >represents the status of the thing we are tracking. There are only a few of >these values. However, I need to be able to read records for certain values >in this column. This must be a common thing to do in tracking systems. > >> What type of locking do you have in that table? >> > > >We have row level locking. > > >