In my previous post, http://groups.google.com/group/google-appengine/browse_thread/thread/c5ab4ec85c5b5525/b8f97af3ddcdcaa1?lnk=gst&q=charming#b8f97af3ddcdcaa1
I said "after I rewrote the custom warm-up servlet, I found it works like a charming". Yes, the warm-up cases really becomes fewer, but now I find it still exits. After some research, I found the behavior of the instance scheduler is uncertain. Here is a case: one resident instance is idle, a new request coming, the idle resident instance will handle it. At the same time, another new request coming, the scheduler wait pending latency time, the idle instance is still not free, so a new instance is started, which needs 20-30 seconds to warmup. After ***N seconds***, the scheduler found the new instance is still not warmed-up and found the resident instance is free now, so the scheduler let the resident instance also hand the second new request. The implementation is not bad, but the strange thing is ***N seconds*** varies from 1 second to 30 seconds. That is why some warm-up requests will be still encountered. btw, I found affer I rewrote the custom warm-up servlet, 99% percent cases of the effort to create a new instance are useless. The new created instance will be shutdown without handling any user requests. -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to google-appengine@googlegroups.com. To unsubscribe from this group, send email to google-appengine+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.