If I am an end user (MRE/BID) and have a long running report request...

If I close the IE window, the DEFAULT agent keeps working to fulfill my request only to deliver the results to a non-existent window.

If an admin kills that agent via the server console, is the request still running on the DBMS server or is it made aware of the cancellation and stops processing? (Oracle if it makes a difference).

- ABT

The request should complete the current query, and then terminate.

But this is a very generic answer, and it may be quite different for each RDBMS.

If an admin kills the agent, the agent will in most cases wait until there is a reply from the dbms it is querying. Only then will it terminate. to ensure a clean termination of the foreign dbms.
This may be different if the admin hard-kills the agent though.

quote:

This may be different if the admin hard-kills the agent though.

Please define 'Hard Kill'? My scenario is using Monitor Agents, via the Admin website (http://server:8121/webconsole?IBIS_page=HOME&contentpage=AGENTS). Right-clicking and selecting 'Kill Agent'. Does that qualify as a Hard Kill?

- ABT

I've been trying to determine the same thing. We have a custom 'please wait' page that has a button for the user to stop the process themselves. The code behind the button looks like

//Begin function stopRequests()
function stopRequests() {
	try {
	location.href='https://servername:port/ibi_apps/Controller?WORP_REQUEST_TYPE=WORP_STOP_SERVER_REQUESTS&';
	this.close();
		}
	catch (err) {
		alert("Error on Stop Request");
		}
}

I can tell that this does stop the request and I can watch the agent go away but I'm not sure if the DBMS is still processing. In our case we use MS SQL Server 2005. I asked the question from support and was told, "A request to gracefully stop the agent is issued. If that doesn’t work, then a hard kill is issued. This is dependent upon the Server Functionality if it supports ending the DBMS process."

Problem is, I don't know how to confirm whether the DBMS process has been stopped or not.
Have any of you been able to determine this?

Thanks!

Dan

I was working with my DBAs on this, I didn't find out much, but here is something I did find out (and maybe a clue of how to proceed for you, Dan).

I picked a large table (8 million rows) and PRINT *'d it. At the same time, my DBA watched the jobs currently running via the Oracle console (actually, I think it was really just a query of current activity from a SYS table).

He was looking for our WF server as the originator and our adapter ID. He said he did see it, but it flashed on and off very quickly.

Meanwhile, my WF Console (View Agents) still was chugging along and my corresponding EDATEMP\TS#### directory was still there with an 800 MB focsort. I let it run a bit longer before killing it, never seeing results in the browser.

What that told me was that intensive operation was intensive only to WF, not the DBMS. In order to (re)perform my test as intended, I'll need to rewrite as something much more DBMS intensive, like a cartesian set or massive join. It is interesting to see the delineation of tasks. Just because you're waiting a long time, it doesn't necessarily give you a hint what you're waiting on.

- ABT