Management and PXE net setup
-
Server
- FOG Version: 1.3.4
- OS: Ubuntu Server 16.04
Hello all,
I need to set up FOG with a management network for web console access and a separate network for PXE booting my machines. DHCP should only be allowed on the PXE net to avoid confrontation with the DHCP server already on the management network. Any assistance that can be provided would be of immense help and would be greatly appreciated.
-
You need two NICs. We can start by seeing what adapters are present in the system and seeing what their configurations are. Please share the output of this command:
ip addr show
Also tell us which subnet you don’t want dhcp served on, and we can give you the lines to add to/etc/dhcp/dhcpd.confso that dhcpd doesn’t serve on the subnet you don’t want it to serve on. -
I am running this on a Dell R610 with multiple 1gb ports built in. I have eno1 active for the management network, so I can access it from my desk, this net should not provide DHCP. The eno2 net should be the PXE net, as it is isolated from everything else and can provide DHCP without conflict. Thank you for your quick response.
Here is the output for ip addr show:
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eno1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
link/ether 14:fe:b5:d1:a2:93 brd ff:ff:ff:ff:ff:ff
inet 10.254.125.216/23 brd 10.254.125.255 scope global eno1
valid_lft forever preferred_lft forever
inet6 fe80::16fe:b5ff:fed1:a293/64 scope link
valid_lft forever preferred_lft forever
3: eno2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
link/ether 14:fe:b5:d1:a2:95 brd ff:ff:ff:ff:ff:ff
inet 192.168.1.1/24 brd 192.168.1.254 scope global eno2
valid_lft forever preferred_lft forever
inet6 fe80::16fe:b5ff:fed1:a295/64 scope link
valid_lft forever preferred_lft forever -
@dws88 What does the storage management page look like for this FOG server?
Also when you setup FOG did you tell it to use/configure en02?
-
@george1421 Honestly, I am a complete noob to FOG, if you couldn’t tell by this thread. So as of right now my knowledge is lacking. I took a look at the storage page and it looks to be configured to use eno1 and its address. I assume that this is wrong and should be set to the interface intended for PXE booting, which is eno2.
-
@dws88 Well that tells me (the page info) that we have a little work ahead (OR) change your perspective and make eno2 your management interface and eno1 your deployment network. That would fix 90% of what we would have to change. For this change just swap your network cables, swap IP addresses on the interfaces and then reboot.
I do have to ask you why are you setting it up this way? (I know the reasons why you should use this configuration, I just want to make sure of your justification because there are a few gotchas when you run an isolated deployment network).
-
@george1421 I am running in an environment that already has an active DHCP server for another system being used by someone else. eno1 is just being used to connect out to our office. If I can prove FOG to be worthy, it could be used to deploy over 130 nodes once in production.
-
@dws88 This is fine having an existing dhcp server. Do you have the ability to change/request dhcp options 66 and 67 on that dhcp server be set to a specific value?
The reason why I ask/pressing this is that you can deploy an image using a dedicated deployment network. That will work for sure. The issue becomes when the client completes imaging usually it will want to connect to AD and register and complete its setup. When you are on an isolated network, reaching AD is impossible unless you take some additional steps. That is why I want to ensure you really want to do this split network.
-
@george1421 Every node has multiple NICs available and most servers are not going to require active directory. The main image I will have this serving will be for storage nodes, along with generic Linux systems, and some other applications as needed.
-
@dws88 OK fair enough.
So to your issue, I do suggest that you change the roles of eno1 and eno2 if possible as well as the IP addresses since FOG is already configured to use eno1 for image deployment. We can reverse this but it will take a few steps.
-
@george1421 If I were to install a clean image would it be possible to have the web UI available on eno1 while the rest of the services act on eno2? That is the primary reason the system was configured the way it was.
-
@dws88 Yes if its a clean install. When you run the installer script for the first time it will ask you for the interface to use (it will default to the first found network interface - eno1). Make sure you pick eno2 there. It will also ask you about the dhcp server, and you only want to bind it to eno2.
If you are going to do a clean install, I also suggest that you jump straight to 1.3.5RC13 or RC14 which ever is current. That (dev) branch has quite a few fixes you will probably need.
-
@george1421 Thank you for your time, you’ve been a great help. I am going to wipe this box and start clean, I’ll check back in a while later and update my progress. It is my opinion, it would be nice to have a simplified interface config page in the web UI.
-
@george1421 Thanks again you’ve been a great help, I went through the install again and was able to PXE boot my first host on the private PXE net.
