It's possible for the oplog entry (and maybe the healthlog entry as well) that dbcheck creates to be too large if the batchStart/batchEnd/lastKeyChecked values are large docs (for example in timeseries bucket collections like here). We should figure out how to handle these cases.