Loss in Throughput after migrating to standalone Nats VM from K8s

[RepoBreather]

Since there is a memory leak issue in Nats k8s cluster as stated in #6004 we tried to move the nats node to 3 node nats cluster running as a systemd process on linux.

we are running 3 node nats jetstream cluster with replication of 3 on 3 different streams.

Node configuration: 2 core 4Gb

Nats-server config:

host: 0.0.0.0
port: 4222
server_name: nats-0

jetstream {
    # Jetstream storage location, limits and encryption
    store_dir: "/var/lib/nats/jetstream"
    max_mem: 1G
    max_file: 20G
}

system_account: SYS

accounts: {
    SYS: {
        users: [
            {user: admin, password: admin}
           ]
    },
    my_user: {
        jetstream: enable
        users: [
            {user: USER, password: PASSWORD}
        ]
    },
}


cluster {
  name: nats_cluster
  listen: 0.0.0.0:6222
  routes = [
    nats-route://nats-1:6222,
    nats-route://nats-2:6222
  ]
}

Our applicatins are in kubernetes but nats nodes are hosted in same network as k8s application nodes.

we are observing lesser throughput for the same test that we ran in k8s nats ( note: all k8s nats node were also having same config of 2core 4Gb without any GOMEMLIMIT or limits on pods)

Test in k8s is giving us around 60 request per second (produce & consume)

same test in nats broker on vm is giving around 48 request per second (produce & consume)

Can anyone pls tell me with an analysis why we are seeing lower numbers in nats hosted in VM?

Also is there a way to stop the nodes from going into OOM state in k8s ?