Charlie Arehart replied with this on [cfaussie]:
>>> "charlie arehart" 10/12/2009 10:58 am >>>
Scott, as far as I can tell, nothing has changed in the CF Enterprise Server
Monitor from 8 to 9. Both it, and FusionReactor, and SeeFusion can all
attempt to kill requests.
Just beware: such an attempt is only that, an "attempt". There's no
guarantee it can be interrupted. If a CF request is in what's called a
native thread (that is, talking to something outside of CF, such as hung
talking to a database, doing a cfhttp, or invoking a web service, etc.) then
none of the tools can interrupt the request.
Yes, that does mean that CF requests can hang for a very long time, indeed
indefinitely in some cases and that you can't really do anything about it.
And no, not even the CF admin timeout setting can change this. It's subject
to the same limitation. It is worth noting that putting a TIMEOUT on CFHTTTP
or CFINVOKE is a good ideas, as sometimes it can allow CF to interrupt
those. We had a discussion of that here last year, if I recall. There's also
a TIMEOUT on CFQUERY too, but it doesn't always work and depends on the
drive version, type, database, and other variables.
Finally, should anyone wonder, all this inability to timeout requests in a
native thread state is not a CF bug. It's a JVM thing. The CF "kill thread"
mechanisms above merely processes ask the JVM to terminate the thread, and
it cannot terminate threads in a "native thread" state.