Hi,
Last night we conducted pre-Hamilton On-Sale Load Testing on our environment to find any bottlenecks or issues. Aside from some not-so-optimized legacy asp.net code in our shop site (it's custom and self hosted), which we're working on, everything else seemed to be able to handle loads up to 500 simultaneous users. We were running perfmon on all the main components of the environment and everything checked out good.
Except one thing: the REST/SOAP API Gateway Server.
No matter if we let 30, 50, or 100 simultaneous simulated users into the test, the PerfMon counter for .NET Application Restart would start ticking up, then ramping up, until we were getting one restart per second.
We did a bunch of research at 1 AM and tried a few things but nothing seemed to work to eliminate those Application Restarts.
Has anyone experienced this in Performance Monitoring? If so, is there a config setting or something we can change?
Here's a screenshot of the perfmon counter, although not from last night. We appreciate any ideas or suggestions. Thanks!
Hi Gerald, What do you see in the Event Logs->System->WAS logs for the API gateway hosts? In many cases, the reason for a system triggered app-pool recycle will be noted there.
Chris
I wonder if you have a nonzero value set for your application pool under Advanced Settings -> Recycling -> Request Limit.