i would guess problems with dnsmasq. It looks like you are trying to put instances on the same network as your host machine which may be difficult to get working if you have an external router or gateway responding to dnsmasq queries. I would check the process list for dnsmasq and make sure that there are exactly two running and that the command line for them is the same and it looks reasonable.

Vish

On Apr 22, 2011, at 3:58 AM, Brian Amedro wrote:

Question #153827 on OpenStack Compute (nova) changed: https://answers.launchpad.net/nova/+q...

Brian Amedro gave more information on the question: I already saw https://answers.launchpad.net/nova/+q... which looks like to be a similar problem, but I am already with FlatDHCPManager. I also tried to connect to the VM through a serial console, but it fails:

==== $ virsh console instance-00000019 ==== Connected to domain instance-00000019 Escape character is ^] error: internal error character device (null) is not using a PTY

==== $ cat /var/lib/nova/instances/instance-00000019/libvirt.xml ==== <domain type="kvm"> <name>instance-00000019</name> <memory>524288</memory> <os> <type>hvm</type> <kernel>/var/lib/nova/instances/instance-00000019/kernel</kernel> <cmdline>root=/dev/vda console=ttyS0</cmdline> </os> <features> <acpi/> </features> <vcpu>1</vcpu> <devices> <disk type="file"> <driver type="qcow2"/> <source file="/var/lib/nova/instances/instance-00000019/disk"/> <target dev="vda" bus="virtio"/> </disk>

   <interface type='bridge'>
       <source bridge='br100'/>
       <mac address='02:16:3e:1b:62:26'/>
       <!--   <model type='virtio'/>  CANT RUN virtio network right now -->
       <filterref filter="nova-instance-instance-00000019-02163e1b6226">
           <parameter name="IP" value="138.96.126.201" />
           <parameter name="DHCPSERVER" value="138.96.126.1" />
       </filterref>
   </interface>
   <!-- The order is significant here.  File must be defined first -->
   <serial type="file">
       <source path='/var/lib/nova/instances/instance-00000019/console.log'/>
       <target port='1'/>
   </serial>

   <console type='pty' tty='/dev/pts/2'>
       <source path='/dev/pts/2'/>
       <target port='0'/>
   </console>

   <serial type='pty'>
       <source path='/dev/pts/2'/>
       <target port='0'/>
   </serial>

   <graphics type='vnc' port='-1' autoport='yes' keymap='en-us' listen='0.0.0.0'/>

</devices> </domain>

You received this question notification because you are a member of Nova Core, which is an answer contact for OpenStack Compute (nova).

Yes that is your problem. It attempts to give the first address in the subnet to your machine and it is conflicting with your external dns server. You could try using a smaller network, such as: 128.96.126.224/27 although your .1 server may still give out addresses and mess things up. Worth a shot. You'll have to delete networks and fixed_ip tables and recreate the smaller range with nova-network network create you should also change: --fixed_range=128.96.224/27 --flat_network_dhcp_start=138.96.126.225

killall dnsmasq iptables -F iptabbles -t nat -F restart nova-network

try again?

Vish

On Apr 22, 2011, at 7:52 AM, Brian Amedro wrote:

Question #153827 on OpenStack Compute (nova) changed: https://answers.launchpad.net/nova/+q...

Status: Answered => Open

Brian Amedro is still having a problem: Hi vish, thanks for the reply !

You're right, I am trying to put instances on the same network as the host machine. This network already has a DHCP server on 138.96.126.1, which give IP for the range 138.96.126.1-199. Also, I cannot modify the config of this server.

Is there any way so the host machine respond to the VM requests instead of the external DHCP server ? Looking at /var/lib/nova/instances/instance-00000019/libvirt.xml, it does not seems to be the case: <parameter name="DHCPSERVER" value="138.96.126.1"/>

Also, and to be complete, even if I put --flat_network_dhcp_start=138.96.126.200 in my nova.config file, started instances was assigned with an IP in the wrong range. Thus, I modified by hand the database, in order to exclude some range by setting the 'reserved' column in the fixed_ips table.

As you asked, here is the list of dnsmasq processes.

nobody 1029 0.0 0.0 21688 1072 ? S 18:16 0:00 dnsmasq --strict-order --bind-interfaces --pid- file=/var/run/libvirt/network/default.pid --conf-file= --except- interface lo --listen-address 192.168.122.1 --dhcp-range 192.168.122.2,192.168.122.254 --dhcp-lease-max=253 --dhcp-no-override

nobody 1417 0.0 0.0 24388 1112 ? S 18:16 0:00 dnsmasq --strict-order --bind-interfaces --conf-file= --domain=novalocal --pid- file=/var/lib/nova/networks/nova-br100.pid --listen-address=138.96.126.1 --except-interface=lo --dhcp-range=138.96.126.200,static,120s --dhcp- hostsfile=/var/lib/nova/networks/nova-br100.conf --dhcp-script=/usr/bin /nova-dhcpbridge --leasefile-ro

root 1418 0.0 0.0 24256 416 ? S 18:16 0:00 dnsmasq --strict-order --bind-interfaces --conf-file= --domain=novalocal --pid- file=/var/lib/nova/networks/nova-br100.pid --listen-address=138.96.126.1 --except-interface=lo --dhcp-range=138.96.126.200,static,120s --dhcp- hostsfile=/var/lib/nova/networks/nova-br100.conf --dhcp-script=/usr/bin /nova-dhcpbridge --leasefile-ro

$ cat /var/lib/nova/networks/nova-br100.conf 02:16:3e:12:e4:ad,i-0000001e.novalocal,138.96.126.200

---

You received this question notification because you are a member of Nova Core, which is an answer contact for OpenStack Compute (nova).

Yes so it isn't actually getting the right ip. I'm not sure if there is a way to get a config like this to work. You may have to switch to using private addressing (like the 10.x default range) and assigning floating ips to give your vms ips on the 138 network.

Vish

On Apr 22, 2011, at 9:48 AM, Brian Amedro wrote:

Question #153827 on OpenStack Compute (nova) changed: https://answers.launchpad.net/nova/+q...

Brian Amedro gave more information on the question: I have some evolution on the error. Now it is a "timed out" error, instead of a "Network is unreachable".

==== euca-get-console-output ==== 2011-04-22 18:24:52,350 - DataSourceEc2.py[WARNING]: waiting for metadata service at http://169.254.169.254/2009-04-04/met... 2011-04-22 18:24:52,353 - DataSourceEc2.py[WARNING]: 18:24:52 [ 1/100]: url error [timed out] 2011-04-22 18:24:55,360 - DataSourceEc2.py[WARNING]: 18:24:55 [ 2/100]: url error [timed out] 2011-04-22 18:24:58,367 - DataSourceEc2.py[WARNING]: 18:24:58 [ 3/100]: url error [timed out] 2011-04-22 18:25:01,373 - DataSourceEc2.py[WARNING]: 18:25:01 [ 4/100]: url error [timed out] 2011-04-22 18:25:02,380 - DataSourceEc2.py[WARNING]: 18:25:02 [ 5/100]: url error [[Errno 113] No route to host] 2011-04-22 18:25:01,373 - DataSourceEc2.py[WARNING]: 18:25:01 [ 4/100]: url error [timed out] Hundred times...

Also, when I ping the instance, I have a response from an other IP : $ nova list +----+-----------+--------+-----------+----------------+ | ID | Name | Status | Public IP | Private IP | +----+-----------+--------+-----------+----------------+ | 35 | Server 35 | ACTIVE | | 138.96.126.227 | +----+-----------+--------+-----------+----------------+

$ ping 138.96.126.227 PING 138.96.126.227 (138.96.126.227) 56(84) bytes of data.

From 138.96.126.225 icmp_seq=1 Destination Host Unreachable From 138.96.126.225 icmp_seq=2 Destination Host Unreachable From 138.96.126.225 icmp_seq=3 Destination Host Unreachable From 138.96.126.225 icmp_seq=5 Destination Host Unreachable

Network was created with:

nova-manage network create 138.96.126.224/27 1 32

---

You received this question notification because you are a member of Nova Core, which is an answer contact for OpenStack Compute (nova).

btw. is it normal that the vnet0 gets an "fe:" mac-address while the vm gets a "02:" mac?