Create mongo backup/restore without indexes

Question

Im wondering how can I create a mongodump/mongorestore without backing up, restoring the indexes?

and how to incrementally restore a mongo db without restoring the indexes?

Answer 1

The mongodump utility creates a binary export of data from MongoDB and saves index definitions and collection options in a metadata.json associated with each database dumped. The index details do not take any significant space in your backup, and will normally be used by mongorestore to re-ensure indexes after each data for each collection is imported from a dump.

If you want to avoid creating any new secondary indexes after the restore completes, mongorestore has a --noIndexRestore option.

Note: The default _id index is required, and always created.

incrementally restore a mongo db without restoring the indexes?

The option for --noIndexRestore applies whether or not you are restoring into an existing database. If you mongorestore into an existing database with indexes using the --noIndexRestore option, no new index definitions will be added but existing indexes will still be updated as data is inserted.

Incremental backup & restore is really a separate question unless you have a simplistic use case: inserting new documents from successive dumps.

As at MongoDB 2.6, the mongorestore utility only inserts documents (i.e. there is no option for updates/upserts). You can use mongorestore to insert multiple dumps into an existing collection, but any documents causing duplicate key exceptions (eg. _id) will be skipped.

I would normally expect that an incremental backup & restore implies taking a delta of changes (all inserts/updates/deletions since a prior backup) and being able to re-apply those to an older copy of the same data. To achieve an incremental backup, you need a history of changes to data, which in MongoDB's case would be provided by the replica set operation log (oplog).