Description:

Enhance monitoring by adding a metric to track the number of queued index builds. This will provide better observability into server behavior when many createIndex requests are queued, helping diagnose issues related to excessive resource consumption and potential memory overhead.

Context:

Currently, concurrent createIndex requests are queued if there are more than maxNumActiveUserIndexBuilds running. These queued requests:

• Consume significant memory resources.
• Count as active connections, even if clients disconnect.

This can lead to:

• Out of Memory (OOM) conditions due to memory usage by queued requests.
• Denial of Service (DOS) scenarios as maxConns may be exhausted because queued requests occupy active connection slots.

Proposed Solution:

Add a new metric to track the number of queued index builds. This metric would help:

• Detect spamming scenarios by monitoring growth trends in queued createIndex requests.
• Diagnose issues related to high memory consumption or connection exhaustion.
• Improve overall visibility into internal behavior when index builds are queued.