Can we have a more detailed view of how concurrency (inbound threads) works within the CA API Gateway?
CA API Gateway 10.x and 11.x
The CA API Gateway has 2 values that define the number of listener threads that can accept connections.
Those are both Cluster-wide properties.
io.httpCoreConcurrency is the default number of threads that are created when gateway.jar is initialized.
The
io.httpMaxConcurrency is the maximum number of threads that will be created for inbound requests once the CoreConcurrency is exceeded and more accept threads need to be generated.
Once io.httpMaxConcurrency is exceeded then threads go into a waiting pool.
Then the tomcat acceptCount property (which defaults to 50) should accept up to 50 connections which are waiting for inbound threads to the gateway.
Ref:
https://tomcat.apache.org/tomcat-9.0-doc/config/http.html
Note: this tomcat property may vary between versions depending on the tomcat version bundled.
Once this Tomcat property is exceeded then you will no longer be able to make any connections to the gateway.