CPU spikes to 100% on Layer7 Gateway MQ listeners after MQ restart

Products

• CA API Gateway

Issue/Introduction

Layer7 Gateway pods in Kubernetes environments may experience sustained 100% CPU usage immediately following a restart of the connected IBM MQ broker. This occurs when inbound MQ listeners attempt to reconnect simultaneously.

Environment

• Layer7 API Gateway 11.x
• Containerized / Kubernetes deployment
• MQ Native Listeners (SSL/TLS enabled)
• Resource allocation: 3 vCPUs or fewer

Cause

Then the MQ broker becomes available after a restart, every configured MQ listener in the Gateway pod attempts to reconnect at the same time. If SSL/TLS is enabled, each listener performs a heavy asymmetric cryptographic handshake. In environments with limited CPU (e.g., 3 vCPUs) and many listeners, this "reconnect storm" saturates the available CPU cores, leading to 100% usage and potential processing delays.

Resolution

Increase Resource Allocation: Increase the CPU and Memory limits for the Gateway pods. For production environments with multiple MQ listeners, a minimum of 4 to 6 vCPUs and 8 GB of RAM is recommended to provide sufficient headroom for burst workloads.