Detailed steps to reproduce the problem?

- create a change stream on a collection that receives ~60k events per hour
- switch between the latest v1 & v2 drivers, observe the memory utilisation over time
- please see attached files of the memory usage, I deployed the v2 driver at 10/18 ~12:00, reverted back to v1 driver at shortly before 10/20 ~00:00.
- please also see attached files of the git diff, I can reproduce this every time when I just switch between the driver version, no other change causes this behaviour.
- please advise what other steps I could do to assist in debugging this.
- on another note: I've observed this months ago when the initial 2.0.0 came out, I then downgraded immediately and waited until 2.3.1 to see if this bug was spotted by somebody else or fixed, but apparently this still exists.

Definition of done: what must be done to consider the task complete?

same stable memory usage as in v2 driver

The exact Go version used, with patch level:

$ go version

1.25.3

The exact version of the Go driver used:

$ go list -m go.mongodb.org/mongo-driver

go.mongodb.org/mongo-driver/v2 v2.3.1

Describe how MongoDB is set up. Local vs Hosted, version, topology, load balanced, etc.

3 node replica set deployed on bare-metal servers, no load-balancers, 8c/16t Ryzen 3700X, 64GB DDR4, 2TB NVMe RAID 1

The operating system and version (e.g. Windows 7, OSX 10.8, ...)

Ubuntu 22.04 LTS

Security Vulnerabilities

none

mongo-go-driver-v1-vs-v2-vs-v1-git-diff.txt