Why is RabbitMQ consuming excessive memory in MetaDefender Secure Storage (MDSS) Kubernetes deployment?

Overview

This article addresses a specific memory exhaustion issue affecting RabbitMQ components within MetaDefender Secure Storage (MDSS) Kubernetes deployments. It details the symptoms of the memory leak and provides a configuration change to the Helm Chart to permanently resolve the instability.

Issue

RabbitMQ v4.1.0 instances might experience continuous memory growth (memory leaks). This is specifically caused by the improper cleanup of "deleted consumer metrics" within the rabbitmq:4.1.0-management-alpine container image.

Symptoms include:

• Steadily climbing RAM usage without release.
• High resource consumption attributed to the Management Plugin.
• Potential Out of Memory (OOM) crashes or pod restarts.

Solution

To resolve this, you will need to switch the container image to the standard Alpine version, which excludes the Management Plugin.

1. Modify the Configuration

Open the values.yaml file used for your MDSS deployment.

2. Update the Image Tag

Locate the RabbitMQ image section and remove the -management suffix.

3. Change From:

image: rabbitmq:4.1.0-management-alpine

Change To:

image: rabbitmq:4.1.0-alpine

4. Apply the Changes

Execute the Helm upgrade command to apply the new configuration:

helm upgrade <release-name> ./<chart-path> -f values.yaml -n <namespace>

5. Monitor

Verify the pod is running the new image and monitor memory usage over the next 24 hours to confirm stability.