mysql 8 NDB php connections drop, cant connect

[System Error Message]

I have a 2 node mysql and 2 node ndb cluster running with 2 dedicated php-fpm servers for the purpose of testing different setups. It is running a php based content website and i am load testing it with a lowly 3000 users at 100 actions per second using jmeter. Everytime i load test i find that even though all the servers in the chain are not maxed in CPU or ram or network, most of the requests fail as the site says that there is a connection error to database.

If i set it to connect to the 2nd node, it just fails to connect outright. The cluster shows that everything is up and running fine, i can connect to any mysql node and make changes or run queries without any issue, but for some reason i can only connect to 1 of the nodes from php while the other simply won't connect. During periods of high loads, both nodes are unavailable and only 1 shows CPU usage.

Most of the tests fail at high loads. at low loads like 10s everything works but CPU is barely used on everything else. The cluster has a lot of resources though this is only development with each database node getting 4 cores, php 8 cores, everything else 2 cores.

I would hate to have to redesign with galera instead as it would take very long not to mention failovers and data replication are a lot harder to handle with gelara even for master slave setup. Right now i can't even get the mysql cluster to pass the most basic step of receiving a decent amount of traffic, it still fails when i stop 1 node.