test_router_add_interface_delete_port_after_failure may fail due to extra port created

[Shih-Hao Li]

In test_router_add_interface_delete_port_after_failure,
when a subnet is created, a neutron DHCP port could be created
if this subnet is DHCP-enabled.

In this case, the test will fail because it only checks if a
router-interface port is deleted and expects no more port in
the system. But it misses the internal DHCP port is still there.

[Shih-Hao Li (shihli)]

A possible solution is posted at

https://review.openstack.org/#/c/354951/

[Nate Johnston (nate-johnston)]

Do you know how frequently this happens in the gate? What error message is thrown when this error is encountered?

[Shih-Hao Li (shihli)]

Sorry I don't how frequently this happens in the gate.
When it happens, assertFalse(plugin.get_ports(context.get_admin_context())) fails
because a DHCP port is still there.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/354951
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=aa52367f78be0941ab7a75ade29a30cd853bbbec
Submitter: Jenkins
Branch: master

commit aa52367f78be0941ab7a75ade29a30cd853bbbec
Author: Shih-Hao Li <email address hidden>
Date: Fri Aug 12 11:59:38 2016 -0700

    Fix test_router_add_interface_delete_port_after_failure

    In test_router_add_interface_delete_port_after_failure,
    when a subnet is created, a neutron DHCP port could be created
    if this subnet is DHCP-enabled.

    In this case, the test will fail because it only checks if a
    router-interface port is deleted and expects no more port in
    the system. But it misses the internal DHCP port is still there.

    The patch here is to explicitly set enable_dhcp=False when
    creating the subnet.

    Change-Id: I4acbf2c83f9145f7b99519ec90934fa862de67f7
    Closes-Bug: #1612894

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/neutron 9.0.0.0b3 development milestone.