Index in memory fluctuation


I am trying to understand how index in memory are managed, but have not found good explanation for that. Could some one help me?

In the Aerospike metrics I see no evictions. Expired objects follow a steady increasing pattern. But I see a sudden drop on number of objects and index in memory, even though not service was restarted in the cluster. What could cause that to happen?


I assume you mean data-in-index. Are you sure nothing restarted? Outside of restarts/expirations/evictions I can only assume someone pushed a delete