Namespace Public Multi-Host Example

This example splits namespace dataplane nodes across two Linux hosts while the controller and Postgres run on a third host. It uses the checked-in config set under examples/ns-public/.

Linux prerequisites

• Three reachable Linux machines:
  • controller host (runs Docker services)
  • VM1 dataplane host
  • VM2 dataplane host
• Controller host: Docker Engine + docker compose.
• VM1/VM2 hosts: sudo, iproute2, and either cargo or a prebuilt nextmini binary.
• Network reachability from VM1/VM2 to controller_addr on TCP port 3000.

Namespace-specific behavior from this example:

• VM1-config.toml and VM2-config.toml both enable:
  • auto_enable_ip_forward = true
  • auto_add_forward_rules = true
  • auto_add_nat = true
• VM2-config.toml sets node_id_offset = 5 so VM2 node IDs do not overlap VM1.
• cleanup.sh removes host veth interfaces; run it only on the dataplane hosts used for this test.

1. Prepare config files on each host

The ns-public example does not use a generator script; edit the three TOML files directly.

On the controller host, confirm examples/ns-public/controller-config.toml matches your node count (default ring with n_nodes = 10).

On VM1 and VM2, update both interface names and controller endpoint in:

• examples/ns-public/VM1-config.toml
• examples/ns-public/VM2-config.toml

By default these files assume public_network_interface = "ens3" and controller_addr = "206.12.95.232:3000".

2. Start controller and Postgres (controller host)

From repository root on the controller host:

docker compose -f examples/ns-public/docker-compose.yml up --build

This launches:

• postgres (with nextmini DB initialized from controller/init.sql)
• controller (reads examples/ns-public/controller-config.toml)

3. Start VM1 dataplane

From repository root on VM1:

cargo build -p nextmini --release
sudo -E env RUST_LOG=info ./target/release/nextmini --config-path examples/ns-public/VM1-config.toml

4. Start VM2 dataplane

From repository root on VM2:

cargo build -p nextmini --release
sudo -E env RUST_LOG=info ./target/release/nextmini --config-path examples/ns-public/VM2-config.toml

Because VM2-config.toml uses node_id_offset = 5, VM2 contributes nodes 6..10 while VM1 contributes 1..5.

5. Verify cluster state

On the controller host, verify all namespace nodes registered:

docker exec postgres psql -U pgusr -d nextmini -c \
  "SELECT COUNT(*) AS nodes_connected, COUNT(DISTINCT node_id) AS unique_nodes FROM nodes;"

For the default configs, expect 10 connected and 10 unique nodes.

You can also watch controller logs:

docker compose -f examples/ns-public/docker-compose.yml logs -f controller

6. Cleanup

On VM1 and VM2:

1. Stop dataplane processes with Ctrl+C.
2. Run veth cleanup:

./examples/ns-public/cleanup.sh

On the controller host, stop controller and Postgres:

docker compose -f examples/ns-public/docker-compose.yml down