This is a weird one. I am trying to run some benchmarks on my API which involves adding 1 million nodes with relationships to the database beforehand. This is done with a direct connection to the database. The Neo4J database is running in an AWS EC2 Instance. When I run my API locally (powershell and docker container), it works perfectly and I am able to quickly add the million nodes (in chunks). As soon as I start my ECS task running my API it no longer works and takes forever to even add the first chunk of nodes. Based on this, it doesn't seem like it's a Neo4J memory configuration problem. Seems like the ECS task is holding transactions or connections somehow different than it does locally.
Might be worth mentioning that I also have shacl validations running too, but those work fine in the first scenario and nothing should be different about the second.
Has anyone seen this sort of behavior before? Any help is appreciated.