When there are a large (>15k) number of docs in the bucket, this test fails. After investigating it for a bit, I think this may be expected as there seems to be some hard limit in one-shot replication that causes the behaviour I'm seeing. Whether the SG behaviour is expected or not, this test should be using a channel filter because there is no need for it to pull every doc in the bucket.
When there are a large (>15k) number of docs in the bucket, this test fails. After investigating it for a bit, I think this may be expected as there seems to be some hard limit in one-shot replication that causes the behaviour I'm seeing.
Whether the SG behaviour is expected or not, this test should be using a channel filter because there is no need for it to pull every doc in the bucket.