Dual Port NIC Config
-
Hi all, CloneDeploy migrant here looking to see if Fog allows me to configure my image server in the way I need it to without having to waste a bunch of time chasing down leads that may not work out. I consider myself a Linux novice and not well versed in advanced networking, but know enough of both to get it done perhaps with a bit of help from a guide or two.
I am looking to set up an image server on a standard Windows 10 PC (Fog would ideally be in a VM on this PC) using a dual port NIC for load balancing purposes. The cable supplying the internet will be connected to the onboard ethernet port and the two ports on the NIC will connect to a switch where the target PC’s will be located. The idea is to only allow so many connections on one IP before the other one takes over so as to not overload a single ethernet port on the server. Both UEFI and Legacy systems will need to be supported. This was done through CloneDeploy via a DHCP Proxy client but I’m not positive how this is done with Fog.
I was able to get CloneDeploy working in my test lab on a local network, but when trying to implement the dual port NIC config I wasn’t able to find enough documentation on it that I didn’t know if it was even possible using that software. As a note I do not have an external DHCP server, so everything would be hosted on the server itself.
I hope this makes sense, I will likely have technical questions at some later point but wanted to see if this was even possible to do before getting to far into things and see if there was possibly some documentation that I could read to help me get there. I look forward to using this software and appreciate the help!
-
The short answer is I would not run FOG on a type-2 hypervisor if you are concerned about performance where you need two nics for load balancing. If you are concerned about speed then run FOG on bare metal.
In using a type-2 hypervisor your load balancing is done at the host OS level since everything on the virtualization layer is done in software.
You might want to start out with what does your environment look like? How many clients do you have? How many sites do you need to support? How many simultaneous systems do you “need” to image at one time? You may find that your disk subsystem is the bottleneck instead of your network, especially if you are running your hypervisor on a desktop class machine. There are many technical decisions you need to make regardless of the deployment tool you pick.
The rest of your wants can be done with FOG or fog supported sub-components like dnsmasq , no problem.
-
@george1421 Thanks for the reply, I appreciate the information! I figured the VM thing might have been a long shot so I’ve already gotten rid of that and set up Debian on bare metal. I’m already noticing this is going to be much easier based on performance alone.
As for the questions regarding the environment, this will be going in our warehouse where at most 15-20 systems will be connected to it at a time, so in the future I will likely be looking for a better, more permanent solution but this will have to do for now. Looks like I’ll have to do some reading up on dnsmasq too, but this looks like a good start.
-
@explosivo98 Just for the records. FOG can be installed with a normal DHCP server (installer asks you about that). That way you don’t need to worry about setting up and configuring dnsmasq manually. While dnsmasq is nice in environments where you have a DHCP server that cannot be modified to add PXE boot information, it’s really not as powerful as standard Linux DHCP server (ISC DHCPd) is! If there is no real need to use dnsmasq I wouldn’t go that route.
-
@explosivo98 Right as Sebastian said FOG can mange dhcp for you. If you can’t/shouldn’t touch your existing dhcp server or you are running a windows dhcp server older than 2012 then dnsmasq is the solution.
I do have a tutorial on dnsmasq that can get you up and running in about 10 minutes: https://forums.fogproject.org/topic/12796/installing-dnsmasq-on-your-fog-server
-
After a few days of testing on a local network just using the onboard ethernet port going to a router this is already a heck of a lot easier than the previous imaging solution that I banged my head against for a month. Bare metal is definitely the way to be and I greatly appreciate the insight. I would have torn what remains of my hair out trying to figure out making that work with Virtualbox.
Going from here, trying to get the final set up the way I described in my original post, it sounds like I need to look into the DHCP configuration to move the imaging to the two additional ethernet ports and off eno1, is that correct? Right now I have eno1, enp2s0f0, and enp2s0f1. eno1 will be supplying internet and I presume the other two will be on different gateways reaching out to the switch where the target PC’s will be connected to. Would this be something I can achieve using the built in tools?
