Revision history [back]

Hi,

we get this problem since a week, so we digged into the cloudbase-init code and landed at networkconfig.py ... [cloudbaseinit/plugins/windows/networkconfig.py] contentpath = networkconfig['contentpath'] contentname = contentpath.rsplit('/', 1)[-1] debiannetworkconf = service.getcontent(content_name)

    LOG.debug('network config content:\n%s' % debian_network_conf)

    # TODO (alexpilotti): implement a proper grammar
    m = re.search(r'iface eth0 inet static\s+'
                  r'address\s+(?P<address>[ˆ\s]+)\s+'
                  r'netmask\s+(?P<netmask>[ˆ\s]+)\s+'
                  r'broadcast\s+(?P<broadcast>[ˆ\s]+)\s+'
                  r'gateway\s+(?P<gateway>[ˆ\s]+)\s+'
                  r'dns\-nameservers\s+(?P<dnsnameservers>[ˆ\r\n]+)\s+',
                  debian_network_conf)
    if not m:
        raise Exception("network_config format not recognized")

    address = m.group('address')
    netmask = m.group('netmask')
    broadcast = m.group('broadcast')
    gateway = m.group('gateway')
    dnsnameservers = m.group('dnsnameservers').strip().split(' ')

    osutils = osutils_factory.get_os_utils()

    network_adapter_name = CONF.network_adapter
    if not network_adapter_name:
        # Get the first available one
        available_adapters = osutils.get_network_adapters()
        if not len(available_adapters):
            raise Exception("No network adapter available")
        network_adapter_name = available_adapters[0]

    LOG.info('Configuring network adapter: \'%s\'' % network_adapter_name)

    reboot_required = osutils.set_static_network_config(
        network_adapter_name, address, netmask, broadcast,
        gateway, dnsnameservers)

... By bootstraping, nova injected a configuration data into this deployed VM like this

auto lo iface lo inet loopback

auto eth0 iface eth0 inet static address 172.16.x.x netmask 255.255.255.0 broadcast 172.16.x.x gateway 172.16.x.x dns-nameservers 172.16.x.x

auto eth1 iface eth1 inet static address 192.168.x.x netmask 255.255.255.0 broadcast 192.168.x.x gateway 192.168.x.x

So, as you can see, the eth1 is not in the eyes of cloudbase-init, so the IP is not set for eth1, cloudbase-init will take per default an IP in range 169.254.x.x for the second NIC via DHCP.

1. Could networkconfig plugin be changed in a better way so that it can handle 2 NICs or general several NICs
2. In getnetworkadapters() method, a SQL statement is invoked. Why is the first net adapter in the return list is chosen for assigning the IP?

Thanks in advance for your help!

Hi,

we get this problem since a week, so we digged into the cloudbase-init code and landed at networkconfig.py ... [cloudbaseinit/plugins/windows/networkconfig.py] contentpath = networkconfig['contentpath'] contentname = contentpath.rsplit('/', 1)[-1] debiannetworkconf = service.getcontent(content_name)

    LOG.debug('network config content:\n%s' % debian_network_conf)

    # TODO (alexpilotti): implement a proper grammar
    m = re.search(r'iface eth0 inet static\s+'
                  r'address\s+(?P<address>[ˆ\s]+)\s+'
                  r'netmask\s+(?P<netmask>[ˆ\s]+)\s+'
                  r'broadcast\s+(?P<broadcast>[ˆ\s]+)\s+'
                  r'gateway\s+(?P<gateway>[ˆ\s]+)\s+'
                  r'dns\-nameservers\s+(?P<dnsnameservers>[ˆ\r\n]+)\s+',
                  debian_network_conf)
    if not m:
        raise Exception("network_config format not recognized")

    address = m.group('address')
    netmask = m.group('netmask')
    broadcast = m.group('broadcast')
    gateway = m.group('gateway')
    dnsnameservers = m.group('dnsnameservers').strip().split(' ')

    osutils = osutils_factory.get_os_utils()

    network_adapter_name = CONF.network_adapter
    if not network_adapter_name:
        # Get the first available one
        available_adapters = osutils.get_network_adapters()
        if not len(available_adapters):
            raise Exception("No network adapter available")
        network_adapter_name = available_adapters[0]

    LOG.info('Configuring network adapter: \'%s\'' % network_adapter_name)

    reboot_required = osutils.set_static_network_config(
        network_adapter_name, address, netmask, broadcast,
        gateway, dnsnameservers)

... By bootstraping, nova injected a configuration data into this deployed VM like this

auto lo iface lo inet loopback

auto eth0 iface eth0 inet static address 172.16.x.x netmask 255.255.255.0 broadcast 172.16.x.x gateway 172.16.x.x dns-nameservers 172.16.x.x

auto eth1 iface eth1 inet static address 192.168.x.x netmask 255.255.255.0 broadcast 192.168.x.x gateway 192.168.x.x

So, as you can see, the eth1 is not in the eyes of cloudbase-init, so the IP is not set for eth1, cloudbase-init will take per default an IP in range 169.254.x.x for the second NIC via DHCP.

1. Could networkconfig plugin be changed in a better way so that it can handle 2 NICs or general several NICs
2. In getnetworkadapters() method, a SQL statement is invoked. Why is the first net adapter in the return list is chosen for assigning the IP?

Thanks in advance for your help!

Hi,

we get this problem since a week, so we digged into the cloudbase-init code and landed at networkconfig.py ... [cloudbaseinit/plugins/windows/networkconfig.py] contentpath = networkconfig['contentpath'] contentname = contentpath.rsplit('/', 1)[-1] debiannetworkconf = service.getcontent(content_name)[cloudbaseinit/plugins/windows/networkconfig.py]

    content_path = network_config['content_path']
    content_name = content_path.rsplit('/', 1)[-1]
    debian_network_conf = service.get_content(content_name)

    LOG.debug('network config content:\n%s' % debian_network_conf)

    # TODO (alexpilotti): implement a proper grammar
    m = re.search(r'iface eth0 inet static\s+'
                  r'address\s+(?P<address>[ˆ\s]+)\s+'
                  r'netmask\s+(?P<netmask>[ˆ\s]+)\s+'
                  r'broadcast\s+(?P<broadcast>[ˆ\s]+)\s+'
                  r'gateway\s+(?P<gateway>[ˆ\s]+)\s+'
                  r'dns\-nameservers\s+(?P<dnsnameservers>[ˆ\r\n]+)\s+',
                  debian_network_conf)
    if not m:
        raise Exception("network_config format not recognized")

    address = m.group('address')
    netmask = m.group('netmask')
    broadcast = m.group('broadcast')
    gateway = m.group('gateway')
    dnsnameservers = m.group('dnsnameservers').strip().split(' ')

    osutils = osutils_factory.get_os_utils()

    network_adapter_name = CONF.network_adapter
    if not network_adapter_name:
        # Get the first available one
        available_adapters = osutils.get_network_adapters()
        if not len(available_adapters):
            raise Exception("No network adapter available")
        network_adapter_name = available_adapters[0]

    LOG.info('Configuring network adapter: \'%s\'' % network_adapter_name)

    reboot_required = osutils.set_static_network_config(
        network_adapter_name, address, netmask, broadcast,
        gateway, dnsnameservers)

... By bootstraping, nova injected a configuration data into this deployed VM like this

auto lo iface lo inet loopback

auto eth0 iface eth0 inet static address 172.16.x.x netmask 255.255.255.0 broadcast 172.16.x.x gateway 172.16.x.x dns-nameservers 172.16.x.x

auto eth1 iface eth1 inet static address 192.168.x.x netmask 255.255.255.0 broadcast 192.168.x.x gateway 192.168.x.x

So, as you can see, the eth1 is not in the eyes of cloudbase-init, so the IP is not set for eth1, cloudbase-init will take per default an IP in range 169.254.x.x for the second NIC via DHCP.

1. Could networkconfig plugin be changed in a better way so that it can handle 2 NICs or general several NICs
2. In getnetworkadapters() method, a SQL statement is invoked. Why is the first net adapter in the return list chosen for assigning the IP?

Thanks in advance for your help!