> I also noticed that even after it came back up, the service > only held up for > so long before it became unresponsive again. I am not having > this problem > with apache 1.3.33.
I'm tellin' ya, you gotta log slow pages and take thread dumps. 1) Take 2 thread dumps when the server is hung up, about 15-30 seconds apart. 2) Find the jrpp stacks containing cfm/cfc template names. They represent the threads in the jvm that were actually executing code at that moment. 3) Note the activity occuring in those jrpp threads with cfms, and read them starting at the top-most cfm in the stack until you get to the top where the jrpp-NN id is located. Reading up, you might see FileTag, meaning CFFILE, or HTTPTag for CFHTTP, or QueryTag for CFQUERY for example. Keep reading up and you might see QueryTag.executeSQL() or something like that, and to continue the example you might see a socketRead0() at the 2nd highest line. In that case what you have is a cfm request running a query and waiting for a query result. If you find that many or all of the jrpp threads are all in database stuff then you have a problem with database connectivity or something of the like. 4) The jrpp threads might show the cfms doing very different things in a single thread dump, so it may not be so obvious. In that case you want to compare the jrpp threads in the two closely timed thread dumps to see if the same jrpp thread is doing the same thing in both thread dumps. I usually pluck out just those jrpps with cfms in them and put them in a separate file, then I analyze that other file for the same jrpp number in both groups from the 2 thread dumps. Then if you see, for example, that jrpp-4 is on foo.cfm:34 (line 34 of foo.cfm) in QueryTag doing a socketRead and you note that jrpp-4 in the second thread dump is exactly the same, then you have just identified a bottleneck thread that is specifically waiting for a database query result. 5) Turn on "log slow pages". Set it low, to 15-30 seconds. Then carefully watch server.log for entries like the following: "Warning","7184","01/31/05","20:14:54",,"Template: G:\users\www\template.cfm, Ran: 258 seconds." 6) Let the server run for a while then make a list of all the templates in server.log that are slow like this. Sort them by frequency and by times. Can you correlate the templates here with those in the thread dumps? 7) Starting troubleshooting those slow or suspect bottleneck templates found in server.log and in the thread dumps and get to work on examining them closely. By far, the most likely reason you're seeing the server hang is because you probably have unexpected bottleneck conditions. Your better off trying to rule this in our out before moving on to other possible suspect problems. -Steve ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| 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:4:192836 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=11502.10531.4 Donations & Support: http://www.houseoffusion.com/tiny.cfm/54
