We have seen in the past where this error can be hit during replication, where the best solution is to resync the node completely. This is a possible data inconsistency issue so we want to add better diagnostics for when this happens since it is near impossible to debug with just log files and without being invasive to the cluster.

The ReplWriterWorker thread can throw this and it will tell us the oplog entry that we were trying to apply when it happened. One idea when we see this error is to perform a snapshot read against the nodes sync source for the document that we were not able to update (the oplog entry should contain an _id for the document) and perform a diff against the nodes local document to detect any inconsistencies.