Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

Have you found a solution ?

I'm also trying to get 14.04 working with Cloud-init v0.7.5. I'm having the same problem.

If I boot 20 VMs with the same 14.04 image, maybe half of them will not ping and I get that error message in the console logs.

What I've found is that soft rebooting the VM will make it ping again.

I also logged to the VM in console, did a "ifdown eth0 && ifup eth0" and the VM started pinging.

Still investigating...

Have you found a solution ?

I'm also trying to get 14.04 working with Cloud-init v0.7.5. I'm having the same problem.

If I boot 20 VMs with the same 14.04 image, maybe half of them will not ping and I get that error message in the console logs.

What I've found is that soft rebooting the VM will make it ping again.

I also logged to the VM in console, did a "ifdown eth0 && ifup eth0" and the VM started pinging.

Still investigating...

Edit: Found what seems to be the problem. I am not using DHCP networks in Openstack, instead relying on Neutron to provide an available address that cloud-init configures when the VM boots.

I use Ubuntu's cloud images (http://cloud-images.ubuntu.com/trusty/current/).

What I learned is that the layout of the interfaces have changed. Instead of just /etc/network/interfaces, you now also have /etc/network/interfaces.d/<interface>.

Now, provided in the image, there is a /etc/network/interfaces.d/eth0 in which you'll find it configured for DHCP:

# The primary network interface
auto eth0
iface eth0 inet dhcp

When cloud-init starts doing it's stuff, it will configure eth0 inside /etc/network/interfaces - as such:

# Injected by Nova on instance boot
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
        address 172.27.58.189
        netmask 255.255.252.0
        broadcast 172.27.59.255
        gateway 172.27.56.1
        dns-nameservers 172.16.131.34

This means you'll end up with two eth0 interfaces configured if the one in interfaces.d is not removed - therefore explaining why 50-60% of my VMs were not pinging, probably running into a race condition.

Have you found a solution ? Found the problem, read my edit below.

I'm also trying to get 14.04 working with Cloud-init v0.7.5. I'm having the same problem.

If I boot 20 VMs with the same 14.04 image, maybe half of them will not ping and I get that error message in the console logs.

What I've found is that soft rebooting the VM will make it ping again.

I also logged to the VM in console, did a "ifdown eth0 && ifup eth0" and the VM started pinging.

Still investigating...

Edit: Found what seems to be the problem. the problem.

I am not using DHCP networks in Openstack, instead relying on Neutron to provide an available address that cloud-init configures when the VM boots.

I use Ubuntu's cloud images (http://cloud-images.ubuntu.com/trusty/current/).

What I learned is that the layout of the interfaces have changed. Instead of just /etc/network/interfaces, you now also have /etc/network/interfaces.d/<interface>.

Now, provided in the image, there is a /etc/network/interfaces.d/eth0 in which you'll find it configured for DHCP:

# The primary network interface
auto eth0
iface eth0 inet dhcp

When cloud-init starts doing it's stuff, it will configure eth0 inside /etc/network/interfaces - as such:

# Injected by Nova on instance boot
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
        address 172.27.58.189
        netmask 255.255.252.0
        broadcast 172.27.59.255
        gateway 172.27.56.1
        dns-nameservers 172.16.131.34

This means you'll end up with two eth0 interfaces configured if the one in interfaces.d is not removed - therefore explaining why 50-60% of my VMs were not pinging, probably running into a race condition.

Have you found a solution ? Found the problem, read my edit below.

I'm also trying to get 14.04 working with Cloud-init v0.7.5. I'm having the same problem.

If I boot 20 VMs with the same 14.04 image, maybe half of them will not ping and I get that error message in the console logs.

What I've found is that soft rebooting the VM will make it ping again.

I also logged to the VM in console, did a "ifdown eth0 && ifup eth0" and the VM started pinging.

Still investigating...

Edit: Found the problem.

I am not using DHCP networks in Openstack, instead relying on Neutron to provide an available address that cloud-init configures when the VM boots.

I use Ubuntu's cloud images (http://cloud-images.ubuntu.com/trusty/current/).

What I learned is that the layout of the interfaces have changed. Instead of just /etc/network/interfaces, you now also have /etc/network/interfaces.d/<interface>.

Now, provided in the image, there is a /etc/network/interfaces.d/eth0 in which you'll find it configured for DHCP:

# The primary network interface
auto eth0
iface eth0 inet dhcp

When cloud-init starts doing it's stuff, it will configure eth0 inside /etc/network/interfaces - as such:

# Injected by Nova on instance boot
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
        address 172.27.58.189
        netmask 255.255.252.0
        broadcast 172.27.59.255
        gateway 172.27.56.1
        dns-nameservers 172.16.131.34

This means you'll end up with two eth0 interfaces configured if the one in interfaces.d is not removed - therefore explaining why 50-60% of my VMs were not pinging, probably running into a race condition.

I filed a bug against cloud-init: https://bugs.launchpad.net/cloud-init/+bug/1315501

Have you found a solution ? Found the problem, read my edit below.

I'm also trying to get 14.04 working with Cloud-init v0.7.5. I'm having the same problem.

If I boot 20 VMs with the same 14.04 image, maybe half of them will not ping and I get that error message in the console logs.

What I've found is that soft rebooting the VM will make it ping again.

I also logged to the VM in console, did a "ifdown eth0 && ifup eth0" and the VM started pinging.

Still investigating...

Edit: Found the problem.

I am not using DHCP networks in Openstack, instead relying on Neutron to provide an available address that cloud-init configures when the VM boots.

I use Ubuntu's cloud images (http://cloud-images.ubuntu.com/trusty/current/).

What I learned is that the layout of the interfaces have changed. Instead of just /etc/network/interfaces, you now also have /etc/network/interfaces.d/<interface>.

Now, provided in the image, there is a /etc/network/interfaces.d/eth0 in which you'll find it configured for DHCP:

# The primary network interface
auto eth0
iface eth0 inet dhcp

When cloud-init starts doing it's stuff, it will configure eth0 inside /etc/network/interfaces - as such:

# Injected by Nova on instance boot
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
        address 172.27.58.189
        netmask 255.255.252.0
        broadcast 172.27.59.255
        gateway 172.27.56.1
        dns-nameservers 172.16.131.34

This means you'll end up with two eth0 interfaces configured if the one in interfaces.d is not removed - therefore explaining why 50-60% of my VMs were not pinging, probably running into a race condition.

I filed a bug against cloud-init: https://bugs.launchpad.net/cloud-init/+bug/1315501