For the last few days I have been using PerfView to look for extensive Large object Heap allocations and monitor GC performance on one of our production web servers(.net 4.7.2 and ASP.NET - MVC 5.2.4), I used the "GC Collect Only", "GC Only", "ETW .NET Alloc" options and possible a few others.
However during and after the trace sessions CPU usage was very high for periods of up to 10 seconds, where all 24 cores were utilized almost 100%. After PerfView session was stopped and the problem persisted I did a 'issreset' and eventually rebooted the webserver. But this did not help, and 15-20 minutes after joining the cluster the server begins to see high CPU utilization again. At this point I created a few dump of the process during high CPU utilization and analyzed them with DebugDiag, this showed that GC was running.
So my question is, what can cause the application to behave this way? Last week the website was running smooth on the that server, as well as 2 others webserver that didn't have problems the last few days. Could there be some kind of lingering trace session that survives restart?