Scheduler starts new dynamic instance instead of using resident ones

[Marcel Manz]

I have set min/max idle instances both to 2. The oldest resident instance serves the most requests, while the secondary waits to accept traffic in case of increased load. So far so good.

Why however the scheduler doesn't make use of that instance and starts spinning up additional dynamic instances, that cause latency because of spin up time:

2011-11-10 07:47:35.552

This request caused a new process to be started for your application, and thus caused your application code to be loaded for the first time. This request may thus take longer and use more CPU than a typical request for your application.

It's not a question about costs, I know that I don't need to pay for any additional instances created above the defined maximum, however it doesn't help my application if latency is caused by unnecessary spin-ups that should be handled by one of the idling resident instances.

Also the

2011-11-10 07:47:35.552 /_ah/warmup 200 2411ms 0kb

warmup time of 2.4 seconds is lower than my specified pending latency of "3.0s - automatic", so there really would be no need for the scheduler to spin up a new instance.

Thanks
Marcel