[Bug] broker cannot be connected

### Search before asking

- [X] I searched in the [issues](https://github.com/apache/pulsar/issues) and found nothing similar.

### Version

server: v2.10.1
client: v2.10.1、cpp、windows

### Minimal reproduce step

We have more than 10000+ clients producing messages at the same time.

Because the client needs to produce very few messages, we create producer->produce message->close producer for every message using the corresponding synchronization interfaces.

But we found that he broker will refuse the connection of the client every few hours until we restart all the brokers. And on and on it goes.

I found noting in the broker's log below except that the broker even could not connect to itself !

(Parts of broker's log and the broker's ip is 11.153.78.236.)

### What did you expect to see?

Clients can produce messages normally.

### What did you see instead?

Clients cannot connect to the broker.

### Anything else?

_No response_

### Are you willing to submit a PR?

- [ ] I'm willing to submit a PR!

No source files or tests are named. Start by reproducing the repeated producer-create, produce, and close cycle with more than 10,000 C++ clients on Windows, then inspect broker logs when connections fail, including the broker's attempted self-connection. Done means clients can continue producing without restarting all brokers.

Written by the indexing model from the issue text.