what could be the reason for reduction of JVM thread count for live traffic.

[ac483h]

let say there are around 100 users/sec traffic available in prod and we were running a test in the same prod env with 1000 user/sec. when we monitor the JVM hazelcast thread counts for all hosts, we could see reduction in the count during our test window. how is this possible? Please find the screenshot attached. the dip in the thread count is the testing time, where more traffic went to the application.

[AntonioSousa]

The JVMs might be reacting to other factors, namely memory. Have you got a lot of GC going on?

[ac483h]

the GC was pretty normal during the duration