We have a couchbase cluster consisting of 3 nodes. Every day we save over 1 million statistical counters and have to delete them once every couple of weeks. We thought of using TTL however, our customers sometime request to save their data longer dynamically, and hence we do not use TTL, as we currently don’t have a method of going back and changing the TTL of millions of counters.
Thus we started using views. We index a view per customer per day and hence have to go over all documents in the view that day and delete them (using bulk operation). This process can take even fifteen minutes (if not longer) and causes a high load on the cluster.
Is there a way to delete all documents under an index in a quick method?