Skip to content

osl_test_netns.md

Latest commit

 

History

History
72 lines (55 loc) · 3.1 KB

osl_test_netns.md

File metadata and controls

72 lines (55 loc) · 3.1 KB
Raw

osl_test_netns

Sets up a veth pair with one end inside a dedicated network namespace, so callers can stand up an isolated, fully bidirectional virtual link for end-to-end testing of network-listening services on a single host.

Unlike osl_fakenic (a dummy interface, which silently drops locally-originated broadcasts), this resource gives the test client a real RX path back to the server, so DHCP OFFERs, ARP replies, and other locally-originated broadcast traffic actually arrive.

The first end (server_interface) is created in the host network namespace so the service under test binds to it normally. The second end (client_interface) is moved into a separate netns whose name is the resource name; test clients invoke ip netns exec <name> ... to talk to the server.

Actions

  • :create: Creates the netns and veth pair, configures both ends (default action)
  • :delete: Removes the veth pair and the netns

Properties

Property Type Default Required Description
netns_name String Resource Name yes Name of the network namespace
server_interface String veth-srv-#{netns_name[0, 6]} Name of the host-side veth peer (max 15 chars per IFNAMSIZ)
server_ip String yes IPv4 address (CIDR) assigned to the server-side interface
client_interface String veth-cli-#{netns_name[0, 6]} Name of the netns-side veth peer (max 15 chars per IFNAMSIZ)
client_ip String yes IPv4 address (CIDR) assigned to the client-side interface
client_mac String MAC address to set on the client-side interface (e.g. for DHCP reservations)

Examples

Create a test netns with explicit interface names and a fixed client MAC:

osl_test_netns 'testclient' do
  server_interface 'veth-srv'
  server_ip        '140.211.166.158/28'
  client_interface 'veth-cli'
  client_ip        '140.211.166.157/28'
  client_mac       '00:1a:4b:a6:a7:c4'
end

Remove the netns and veth pair:

osl_test_netns 'testclient' do
  action :delete
end

Using the netns from InSpec or scripts

# Run a command inside the netns:
ip netns exec testclient ping -c1 140.211.166.158

# Send a DHCPDISCOVER via the client interface:
sudo ip netns exec testclient /usr/local/bin/dhcp4-probe.py --iface veth-cli

Notes

  • veth is in the upstream kernel without a separate module load, so no kernel_module resource is needed (unlike osl_fakenic's dummy load).
  • Inside containers (dokken), ip netns needs CAP_SYS_ADMIN and a writable /var/run/netns. The repo's kitchen.dokken.yml runs with privileged: true, which provides both.