Packstack Stein - Cannot Detect Tunnel Interface

asked 2019-07-09 08:44:33 -0500

toddleish gravatar image

updated 2019-07-11 07:30:39 -0500

Hello, I am attempting to use Packstack with an answer file to install Openstack on two nodes -- a controller/network node and a compute node. I installed Packstack for Stein new today.

When I run packstack with the answer file, it gets to the step: "Preparing Neutron API entries" and then errors out. The error reads: ERROR : "Couldn't detect ipaddress of interface ens2f1 on node 10.57.38.123"

The controller node management IP is 10.57.38.123 (on bond0). I am trying to use ens2f1 for the OVN tunnel. It's IP address is 10.57.37.123 (verified on the server).

I will post the networking part of the answer file and the setup log with failure below. The entire answer file won't post correctly.


# Password to use for OpenStack Networking (neutron) to authenticate
# with the Identity service.
CONFIG_NEUTRON_KS_PW=##

# The password to use for OpenStack Networking to access the
# database.
CONFIG_NEUTRON_DB_PW=##

# The name of the Open vSwitch bridge (or empty for linuxbridge) for
# the OpenStack Networking L3 agent to use for external  traffic.
# Specify 'provider' if you intend to use a provider network to handle
# external traffic.
CONFIG_NEUTRON_L3_EXT_BRIDGE=br-ex

# Password for the OpenStack Networking metadata agent.
CONFIG_NEUTRON_METADATA_PW=91f9fac7af944e61

# Specify 'y' to install OpenStack Networking's Load-Balancing-
# as-a-Service (LBaaS). ['y', 'n']
CONFIG_LBAAS_INSTALL=n

# Specify 'y' to install OpenStack Networking's L3 Metering agent
# ['y', 'n']
CONFIG_NEUTRON_METERING_AGENT_INSTALL=y

# Specify 'y' to configure OpenStack Networking's Firewall-
# as-a-Service (FWaaS). ['y', 'n']
CONFIG_NEUTRON_FWAAS=n

# Specify 'y' to configure OpenStack Networking's VPN-as-a-Service
# (VPNaaS). ['y', 'n']
CONFIG_NEUTRON_VPNAAS=n

# Comma-separated list of network-type driver entry points to be
# loaded from the neutron.ml2.type_drivers namespace. ['local',
# 'flat', 'vlan', 'gre', 'vxlan', 'geneve']
CONFIG_NEUTRON_ML2_TYPE_DRIVERS=vxlan,flat,geneve

# Comma-separated, ordered list of network types to allocate as
# tenant networks. The 'local' value is only useful for single-box
# testing and provides no connectivity between hosts. ['local',
# 'vlan', 'gre', 'vxlan', 'geneve']
CONFIG_NEUTRON_ML2_TENANT_NETWORK_TYPES=vxlan,geneve

# Comma-separated ordered list of networking mechanism driver entry
# points to be loaded from the neutron.ml2.mechanism_drivers
# namespace. ['logger', 'test', 'linuxbridge', 'openvswitch',
# 'hyperv', 'ncs', 'arista', 'cisco_nexus', 'mlnx', 'l2population',
# 'sriovnicswitch', 'ovn']
CONFIG_NEUTRON_ML2_MECHANISM_DRIVERS=ovn

# Comma-separated list of physical_network names with which flat
# networks can be created. Use * to allow flat networks with arbitrary
# physical_network names.
CONFIG_NEUTRON_ML2_FLAT_NETWORKS=*

# Comma-separated list of <physical_network>:<vlan_min>:<vlan_max> or
# <physical_network> specifying physical_network names usable for VLAN
# provider and tenant networks, as well as ranges of VLAN tags on each
# available for allocation to tenant networks.
CONFIG_NEUTRON_ML2_VLAN_RANGES=

# Comma-separated list of <tun_min>:<tun_max> tuples enumerating
# ranges of GRE tunnel IDs that are available for tenant-network
# allocation. A tuple must be an array with tun_max +1 - tun_min >
# 1000000.
CONFIG_NEUTRON_ML2_TUNNEL_ID_RANGES=

# Comma-separated list of addresses for VXLAN multicast group. If
# left empty, disables VXLAN from sending allocate broadcast traffic
# (disables multicast VXLAN mode). Should be a Multicast IP (v4 or v6)
# address.
CONFIG_NEUTRON_ML2_VXLAN_GROUP=

# Comma-separated list of <vni_min>:<vni_max> tuples enumerating
# ranges of VXLAN VNI IDs that are available for tenant network
# allocation. Minimum value is 0 and maximum value is 16777215.
CONFIG_NEUTRON_ML2_VNI_RANGES=10:100

# Name ...
(more)
edit retag flag offensive close merge delete

Comments

I had a quick look at the source without fully understanding where the error might come from.

One detail that caught my attention: You have the VXLAN type driver, and use it as the default tenant network type, though OVN requires Geneve only. Perhaps that confuses the code?

Bernd Bausch gravatar imageBernd Bausch ( 2019-07-10 04:24:25 -0500 )edit

If you can live with OVS instead of OVN, see how I was successful setting up Devstack.

Bernd Bausch gravatar imageBernd Bausch ( 2019-07-10 04:25:27 -0500 )edit

Thank you for your reply. I removed VXLAN and used Geneve as the driver, but I still got the same result when I re-ran Packstack: ERROR : "Couldn't detect ipaddress of interface ens2f1 on node 10.57.38.123" Do you think this could be a bug?

toddleish gravatar imagetoddleish ( 2019-07-10 13:03:27 -0500 )edit

A bug is always a possibility. In your position, I would do one of the following:

  • Set up the simplest possible OVN config and gradually adapt it until it breaks
  • Dig around in the source code until I understand where the error is from
  • Not use OVN
Bernd Bausch gravatar imageBernd Bausch ( 2019-07-11 07:30:00 -0500 )edit

For now, I added RDO to the question tags, in the hope that someone from the Packstack team replies to your question.

Bernd Bausch gravatar imageBernd Bausch ( 2019-07-11 07:31:19 -0500 )edit