You can add that setting to existing indexes. New buckets will be index cluster buckets and replicate. The previous buckets will remain as standalone buckets, will be searchable, but will not replicate. As an example:
DB Directory for an index on an index cluster member, index is not configured with repFactor=auto
Oct 4 10:38 db_1538594348_1538594348_0
After adding 'repFactor=auto
Oct 4 10:38 db_1538594348_1538594348_0 Oct 4 10:42 db_1538594348_1538594348_1_81D21F8E-64D7-4ACB-962A-3CB331958491
I don't see any problems.
I would run first on the one of the indexers -
./splunk btool indexes list --debug
That would show you the entire set-up of your indexes. By default for indexes where
repFactor was not defined you should see
repFactor = 0.
From Managing Indexers and Clusters of Indexers
and for the sake of completeness
I guess you can assign
repFactor = auto to existing indexes but I think only new buckets will be replicated to other indexers. I'll recommend to test this in test environment first.