Kubernetes Events TTL configuration

We have one AKS on Azure Stack Hub, which was provisioned via AKS engine.
AKS version: 1.24.9
AKS engine version: v0.75.3.

Recently we noticed that there were lots of etcd server timeout.
We find that a huge number of events are stored in ETCD. Around 1 million.
Normally K8S was supposed to store the latest 1 or 2 hours event. But in our AKS we can find some events which older than 1 month.

So I would like to understand more details about:

- What is tthe default --event-ttl in AKS engine?
- How can we override this settings for an existing cluster?
- How to override this for a new provisioning?