What is the likelihood of encountering "missing messages" in message bank?
My team works on implementing an Interoperability solution utilizing InterSystems Kubernetes Operator on Red Hat OpenShift container platform.
We are trying to determine how many messages we can process in any given time. We have a Feeder app running in 10 containers sending 50k messages each to a load balancer all beginning at the same time.
Messages are received via HTTPS protocol by webgateway containers.
Interoperability production runs in compute pods with persistent data, journals, and WIJ volumes.
We implemented Horizontal Pod Autoscaler to scale compute pods when CPU utilization is high.
We utilize Enterprise Message Bank to have one place to find any message processed by any compute.
We observe the queue for Message Bank operation grows quite large in compute pods.
Sometimes the Autoscaler scales computes down while they are still processing.
How likely is it Messages do not show up in Message Bank if they have been partially or completed processed in compute pods?
Any messages in queues on compute pods which are shutdown will not be processed until the compute pods is started again.