Hi,
I am a new user of Aerospike (Enterprise version). I am running asbench tests for a long period of time to benchmark the performance and durability of different SSD disk hardwares. So far, I have a single Aerospike node with replication factor of 1. The namespace being tests has 2 SSD partitions, each with 375GiB space. This is my test environment.
My quesiton is about insert operations, how does Aerospike handle repeated insertions on the same set of keys (same number of keys and same start key)? The reason asking is that I notice this insertion eats up the available disk space, where the metric aerospike.namespace.disk.available goes down gradually while device usage bytes don’t increase, and eventually the insertion fails with error AEROSPIKE_ERR_SERVER_FULL even though the used device bytes is ~500GiB (total disk space is 750GiB). Could someone share some insights on what’s going on behind the scene? And how to make the remaining ~250GiB blocks usable to Aerospike again?
P.S. the Aerospike version used here is E-6.2.0.32 and there’s no special reason for running repeated insertions on the same set of keys. I just try to understand how Aerospike handles it internally.
Thanks!