1

This environment consist of virtual machines running in Ubuntu. The juju server is running in Ubuntu 14.04. I have 4 virtual machines running in ppcel Ubuntu 15.04. The network was functioning properly when the deployment began. As soon as neutron was installed on the bootstrap host the network failed. The network only failed on the bootstrap host the, network on all others are fine. The charm deployment failed due to the failure of the network.

I have checked network configuration and failed to find any problem. Does anyone have any idea what is causing this failure.

2 Answers 2

1

Prolog

I work with Deva at IBM on a proof of concept that brought this issue up.
We were able to make the juju OpenStack 40 charm get by this issue. Note that this pertains to using the charm for ppc64el deployment.

Charm: https://jujucharms.com/u/openstack-charmers-next/openstack-base/40

The documentation suggests "Two cabled network ports on eth0 and eth1 (see below)." This is a bit skimpy on what is really needed. OpenStack setup for neutron (networking) is a bit complicated. See the OpenStack documentation link provided below. OpenStack would like to have network interface 1 for management use (eth0). It would like to have network interface 2 for use of neutron based on the eth1 definition in the charms used by the openstack 40 bundle. Now we originally had eth1 as the external. But since neutron wants to use eth1 per this bundle setup it then intends to use eth1. When this occurs having the external ip address on eth1 ends up being ignored and the server looses external connectivity.

What we needed to do was follow OpenStack documentation to use eth0 as the management interface, we used private network. We set up eth1 as the way OpenStack suggests. See section "To configure networking:" in below OpenStack URL. eth1 as

# The external network interface
auto INTERFACE_NAME
iface INTERFACE_NAME inet manual
    up ip link set dev $IFACE up
    down ip link set dev $IFACE down

Now this has eth0 and eth1 setup per the openstack 40 bundle documentation of having two network interfaces. You still have the problem that you need to define the external ip address so adding a bridge fixes that problem. Something list so:

auto br0 
iface br0 inet static
       address 9.3.80.444
       netmask 255.255.255.0
       network 9.3.80.0
       broadcast 9.3.80.255
       gateway 9.3.80.1
       bridge_ports eth1
       bridge_fd  9
       bridge_hello 2
       bridge_maxage 12
       bridge_stp off

So the way we got our initial manual environment server (vm2) to get external access is by fixing /etc/network/interfaces as shown below.

OpenStack Install Instructions

Need to put https: in front of following URL: //docs.openstack.org/kilo/install-guide/install/apt/content/ch_basic_environment.html#basics-networking

/etc/network/interfaces

auto lo

auto eth0
iface eth0 inet static
   address 192.168.93.76
   netmask 255.255.255.0

auto eth1
iface eth1 inet manual
      up ip link set dev $IFACE up
      down ip link set dev $IFACE down

auto br0 
iface br0 inet static
       address 9.3.80.444
       netmask 255.255.255.0
       network 9.3.80.0
       broadcast 9.3.80.255
       gateway 9.3.80.1
       bridge_ports eth1
       bridge_fd  9
       bridge_hello 2
       bridge_maxage 12
       bridge_stp off

#dns-* options are implemented by the resolvconf package, if installed
        dns-search aus.stglabs.ibm.com
        dns-nameservers 9.3.1.200 9.0.128.50
1
0

"The network was functioning properly when the deployment began. As soon as neutron was installed on the bootstrap host the network failed. The network only failed on the bootstrap host, the network on all others are fine..."

Not exactly sure what you're trying to say here when you suggest "a network failure" in your title and you then suggest that only one computer is affected. I tend to think of network failures as "all or most of the computers can't talk to each other". Anyway, Neutron is the networking service part of OpenStack in case you haven't figured this out. It's responsible for managing routing between what's usually your private network and the more public side of things. Later, when you "expose" a service Neutron then is issuing a public IP address to that service so that it can be seen by the outside world.

Why does it fail at the Neutron step (from a networking standpoint)? The most likely cause would be improperly setting up the network configuration.

Revisit the network settings especially default gateway and netmask. In MAAS this will be in the Cluster and Network tabs. Make sure to review every setting and drill down into the Ethernet adapters themselves on the cluster controller.

3
  • Thanks you for your reply. The problem is with the bootstrap host only.
    – D.Walksfar
    Oct 14, 2015 at 14:53
  • This issue was caused by incomplete configuration. Neutron requires a bridge that we did not have setup. Add the following to /etc/network/interfaces auto eth1 iface eth1 inet manual up ip link set dev $IFACE up down ip link set dev $IFACE down auto br0 iface br0 inet static address x.x.x.x netmask 255.255.255.0 network x.x.x.x broadcast x.x.x.255 gateway x.x.x.x bridge_ports eth1 bridge_fd 9 bridge_hello 2 bridge_maxage 12 bridge_stp off
    – D.Walksfar
    Oct 14, 2015 at 15:06
  • Well done then. It's funny how automated all of this is at times and how--in your case--it was necessary to drop down into a shell prompt and manipulate a configuration file for everything to behave. Oct 15, 2015 at 19:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .