The issues Jennifer and Steven brought up were pretty much spot on. If you have something hanging a request and you pile on simultaneous requests that will do nothing to address the bottleneck.
I would do things a little differently but the same basic principles apply. First of all, I'm going to assume this is a crisis situation, and you need at least a resemblence of stability first and diagnosis/correction can follow, so long as you stop the system from dying. First, REDUCE the number of simultaneous threads to as low of a number as you can, with one being the minimum and the most stable setting. Just because its more likely to be stable doesn't mean it will perform 'well'. This will indeed introduce a performance bottleneck, but its meant to. When in crisis mode this is often the way to stabilize a server in an emergency, letting it run long enough to work towards a fix while users get to keep using it. Be careful (i.e. watch it while this is going on to make sure you have not created a new problem). Still, only the busiest box will time threads out that are queued as the result of this. Next enable a CF restart after X threads have hung, where the restart value is one LESS than the value set in the first step above. Set the hung thread count higher than the simultaneous request count and it will never fire. In an emergency situation I would set this to one (1) thread. Again, pay close attention to what is going on because the only thing worse than a CF box that keeps dying is one that restarts itself every 15 seconds. Yes, this is aggressive but if you get even one hung thread chances are by the time the second one arrives CF will be done for and no longer able to restart itself without human intervention. Next, enable the tracking of long-running page requests. If the above two steps have stabilized your server enough so it can run (bearing in mind that each or the two in tandem can make things worse for a heavy-traffic box), this remains an excellent tool for debugging, despite its noted weaknesses. lastly, be sure to check your mail spool file. What is being described sounds a lot like that problem. I ran into it on a CF5 box a few months back and CF was dying and as soon as it got started up again it instantly croaked until I remembered this problem, checked for it and deleted the offending file (you have to stop the CF service first or the file will be locked and is undeletable. HtH, -- --mattRobertson-- Janitor, MSB Web Systems mysecretbase.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Logware (www.logware.us): a new and convenient web-based time tracking application. Start tracking and documenting hours spent on a project or with a client with Logware today. Try it for free with a 15 day trial account. http://www.houseoffusion.com/banners/view.cfm?bannerid=67 Message: http://www.houseoffusion.com/lists.cfm/link=i:4:211196 Archives: http://www.houseoffusion.com/cf_lists/threads.cfm/4 Subscription: http://www.houseoffusion.com/lists.cfm/link=s:4 Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=89.70.4 Donations & Support: http://www.houseoffusion.com/tiny.cfm/54
