Anderson,
if you are doing this insertion of data into SQL Server for Mongodb in a single transaction, you already ensure that the data is equal, ie only enter into Mongodb if the data has been successfully entered into the relational database and then do the commit transaction. You can do this by using a Transactionscope.
If you are doing this operation asynchronously, that is, you insert it into the SQL database, play in a queue for example and then consume it from that queue to play in Mongodb.
You can use a standard to work with messaging called Outbox, adapted to your problem. This way you will have a table of everything that came out to your synchronization queue, you can record that the data has not been consumed successfully, the process that will take the data and enter in Mongodb is who will signal that the data consumption has been done.
So you have a table that marks the output of the data from your SQL base to Mongodb, as if it were an "output box". One benefit of this approach is that you can see what hasn’t been synchronized yet and you can have an application just looking at this sync table, which will tell you what has been synchronized or not, giving you the benefit of making a new attempt to synchronize the data, if you make a structure for it.
"creating a log table in both banks" this would be an alternative, now without knowing the banks, tables, infra and more other factors, any response will be based on opinions and will not help much
– Ricardo Pontual
Well Ricardo had not seen his questioning before, the CQRS architecture standard is in it are already synchronized the two banks when there is data change, so there is already a synchronization, but as they are distinct servers there may be a failure to insert in mongodb, if there is a fault I was thinking of sending to a message service, when reading the queue it would update the mongodb and the Entity framework allows you to create a log in this way knowing which tables have been changed https://stackoverflow.com/questions/25245351/entity-framework-6-Audit-log
– Anderson Tavares