On investigation flaky test in our codebase I found strange issue: sometimes I got change stream update event with empty fullDocument field for existing document.

I reproduced this behaviour by looping with same collection:

• drop collection;
• shard collection;
• open change stream;
• run background thread which add ~2000 records and then update ~100 records;
• read change stream and check then update event has non-empty fullDocument.

Environment details:

• one shard from single-node replicaset;
• single-node configuration replicaset;
• 3 mongos.

On my computer, a spike of errors occurs in the first 10-30 iterations. Then the probability of errors greatly decreases.

I wrote a simple program with Golang for reproducing this issue (main.go). This program uses go.mongodb.org/mongo-driver v1.3.4 for MongoDB access but originally this issue was reproduced with github.com/globalsign/mgo fork.