The Question and Answer Board

[Bob Zumbrunnen]

From what we're finding in searches, similar but the opposite.

What little info there is suggests that the problem is not doing an rs.close and set rs=nothing each time an rs isn't needed anymore.

Since rs is the handle I usually use, I did add something today to the last include file every page calls where I see if rs is a recordset, then if it is, I check to see if it's open and if it is, I close it and set it to nothing.

I did the same thing with my_conn which is the handle I almost always use for the db connection.

I didn't put it into production, though, because I found that some ASP's were doing a my_conn.close and set my_conn=nothing *after* the last include file is called. Which made them crash.

So I programmatically eliminated every my_conn.close from all of the ASP's in the development environment today and Matt's going to run through it tonight to make sure everything works and copy it into production if it does.

This by itself could very well solve the problem.

However, tomorrow I'm going to (again, programmatically) accumulate all of the recordset names that're ever used on the site (we expect it's between 10 and 20), kill their close statements in the ASP's, and do the same testing/closing in the final include. Matt will test the results tomorrow night and put everything into production if it's working.

By Friday we should know for certain that at the end of every page there are no open recordsets or connections.

If the site still continues to crash unexpectedly, I'm going to be really stumped, as all of our searching yielded only this as a possible cause aside from the Microsoft explanations which we know 100% certain aren't issues here. Specifically, CPU over-utilization or connection threshhold being exceeded.