we are running into this similar issue, and this exception from mongo is causing 500 errors on our Api.
What I want to know is, where exactly is the problem. Upgrading the entire cluster is a more time consuming task.
Is the issue with the router ( mongos ) or is it with the configserver ( mongod ) is it with the latest Java driver?

We want to upgrade the entire cluster, but due to time constraints, I was hoping to get a response about which part of the cluster is broken, and if we upgrade only that, will it solve the problem. Also, it would be great to get a definitive answer about the fix version?
A quick response would be appreciated because its affecting our service.

Thanks.