VirtualIron.com | Join | Contact Us

Blog


Welcome to the Virtual Iron Forums!
Here you'll find information to help you get the maximum value from your Virtual Iron software.

Community
Decrease font size
Increase font size
Topic Title: Can't use management LAN for VS
Topic Summary: If I try to assign a VNIC of a VS to LAN of the management I loose connectivity to the node
Created On: 01/08/2007 06:11 AM
Linear : Threading : Single : Branch
Topic Tools Topic Tools
View topic in raw text format. Print this topic.
 01/08/2007 06:11 AM
User is offline View Users Profile Print this message


fcoudert
New User

Posts: 9
Joined: 01/05/2007

Hi,

I selected "shared network" during installation of the management server, but if I try to assign a VNIC to the LAN of the management I loose connectivity to the node (see tcpdump output below). It looks the arp layer is broken.

My setup is a PE 1950 with 3 ethernets interfaces :
* eth0 not in use,
* eth1 in use on a production LAN,
* eth2 in use on another production LAN also shared for management/PXE boot. (in fact eth2 is the first interface which initiate PXE boot, but it appears as eth2 on Linux système)

I rebooted the node. It appears alive on management but somme errors occurs when I tried to start a VM on it (XML parsing error). I must reboot it again, but from the management console, to get all back in normal operationnal state.

Tcpdump output between management station and the node, continuously repeating :
15:48:04.675130 arp who-has 192.168.239.40 tell 192.168.239.44
15:48:04.675140 arp reply 192.168.239.40 is-at aa:00:00:ed:00:03
15:48:04.689999 IP 192.168.239.40.38133 > 192.168.239.44.777: S 2452311088:2452311088(0) win 5840 0,nop,wscale 4>

Edited: 01/08/2007 at 09:45 AM by fcoudert
 01/08/2007 02:26 PM
User is offline View Users Profile Print this message


dohair
Virtual Iron

Posts: 1
Joined: 01/08/2007

Hi, thanks for evaluating Virtual Iron. The problem you are describing here is caused by the fact that you are using the shared network configuration, but also using the management server DHCP/TFTP services to PXE-boot the compute nodes. This is not a supported configuration and can cause problems especially if there is already another DHCP server on the shared network. PXE-booting the compute nodes is highly recommended but it requires a completely separate Ethernet switch or VLAN for a Virtual Iron private management network. This requires a second Ethernet port on both the management node and all compute nodes. The general procedure is as follows (assuming a Linux mgmt node):
    Connect the management node and all compute nodes to the private network switch or VLAN. The typical wiring pattern is:
      Management node: Ethernet port 1 is public, port 2 is private.
      Compute nodes: Ethernet port 1 is private, port 2 is public.
    If you only have a single compute node, you may use a crossover cable connecting Port 2 of the management node with Port 1 of the compute node.
    On the management node, configure a second Ethernet device (i.e., eth1) with a static IP address outside the range of the first device (i.e., eth0). I usually use IP=192.168.2.1 with a netmask of 255.255.255.0, no gateway or nameserver.
    Uninstall Virtual Iron, then re-install, choosing the "separate management network" option. Be sure to assign the correct Ethernet devices to the public and management networks (choose the correct IP address from the drop-down menus).
    Also during the re-install, enable the DHCP server on the management network. This should be the default choice.
    After the re-install, restart the compute nodes. They should automatically PXE-boot the virtualization layer from the management node and be automatically discovered within the Virtual Iron management console. From this point you can proceed with the configuration of your virtual infrastructure.
PXE-booting the compute nodes as described above is highly recommended for all Virtual Iron installations. However we also support booting the compute nodes from a CD-ROM burned from the VirtualizationServices.iso we include in the Virtual Iron management server installation. In this case, you would select the shared network and no DHCP server during management server installation. In order for this to work, the shared network must already have DHCP services and you must manually discover the compute node after it boots. The separate public/private networks with DHCP services on the management node is highly recommended for all Virtual Iron installations. The CD-ROM boot with shared network configuration should only be used when the management node and/or compute nodes don't have a second Ethernet port for a private management network.

-------------------------
Dave O'Hair Virtual Iron Technical Support
 01/08/2007 06:36 PM
User is offline View Users Profile Print this message


fcoudert
New User

Posts: 9
Joined: 01/05/2007

Thank's dohair for this clarification.

I need the three interfaces for three different production LANs, this is why a tried to share one with management. But after all I guess doing a dedicated LAN for management is a more secure solution. I'll add another ethernet card.
Statistics
222 users are registered to the Community forum.
There are currently 0 users logged in.

FuseTalk Standard Edition - © 1999-2007 FuseTalk Inc. All rights reserved.


Copyright © 2003-2007 Virtual Iron Software, Inc. | Privacy Statement | Terms of Use | Site Map