We have Kubernetes cluster in which we have deployed a Redis instance using Helm chart chart=redis-9.4.3 , Disk is 8gb for persistence via AOF way.
Image : docker.io/bitnami/redis:5.0.5-debian-9-r169
As the consumption was increasing day by day we opted to move to Azure cache for Redis plus other reason as well for migration.
We had 39 clients and all moved to azure cache for redis. There were 50k keys in the db which were attached to various clients but not being used and TTL was -1 (no expiry)
As we moved to Azure, After 24 hours of this operation we noticed a sudden drop in OLD REDIS usage, and all the keys got deleted.
There was not a single key in the DB.
I cross-checked and there was no eviction of keys and settings are as:
maxmemory:0
maxmemory_human:0B
maxmemory_policy:noeviction
The graph for memory change is as : 
I tried searching for the same in the documentation but there seems to be no explanation regarding this behavior.
The Possibility of flushing the DB is 0% as those commands are turned off from the config map.
What can be the possible reason behind it?