Features questions: Fog with distributed architecture
I need to deploy linux on clients located in 4 different sites.
For now, i’m using a PXE/TFTP/DHCP server by site. Each client boot over network on the PXE server of his site, get a debian netinstall and run a preseed script.
The problem is, it’s ok for now with 4 sites to manage, but it will be a nightmare with some more.
I found the FOG project very interesting, but i have some questions:
Networking (source https://wiki.fogproject.org/wiki/index.php?title=Fog_deployment_nodes)
In my case, i need a “master fog” in my “master site” and one “node fog” on all other sites.
I understand that the master fog communicate with all nodes. But i’m not sure about clients, do they communicate with their node or directly with the master fog? Because our site-to-mastersite link is really poor.
Actually, a client install debian from a netinstall (to have a basic system) and then execute a preseed script (for all specifics parameters). Is it possible to keep this method with Fog? Or is there an other way?
Thank you for the reply
@martinm Using postdownload scripts might be an option (depending on what things you wanna change). Just give it a try, it’s not hard. Search the forum and you will find an example. Regarding IP and hostname change you can look into this or give it a couple more days till the new client will be available on linux (which would give you even more options like running snapin tasks on the clients!).
@martinm v0.10.0 of the client (osx and linux support) currently has a release candidate being tested by the development team of FOG.
Thank you Sebastian for the advice. Actually, in my case, imaging (not installing) can be a solution.
I just need specific values for each hostgroup (like specific site parameters) or for each host (for chaning the ip address, hostname…).
But as I know, it can’t be done with Fog for now, because there is no linux client yet.
I already found the FAI project, but i can’t deal with remote sites with this solution. If you know other projects, please let me know!
Anyway, thank you very much for your help and for the amazing job you’ve done on the FOG Project!
EDIT: I found here that we can execute post-deploy script with environment variable. It can be a solution, but i can’t find any documentation on the subject. Any help?
As Wayne already said, FOG is made for imaging - not installing. You may be able to do this via a custom menu entry in FOG trunk but there are a lot of other projects focusing on just that, like FAI…
Thank you very much for the reply Wayne,
I have to say that this plugin is the answer to all my fears about this project.
And also, you’re right, 1 min to deploy a client is huge compared to a netinstall, even with apt-cacher-ng it takes about 30 min to do everything i need…
I’ll give it a try !
Ps: Do you know what “in the future” means for the linux FOG client? Weeks, months, years?
This post is deleted!
networking - Depends on how you set it up. In fog-trunk, there is the location plugin. It can be set so nodes serve the kernel and init to hosts, and hosts can grab tftp from the nodes too. Sounds like you should use the location plugin due to your slow WAN link.
pressed - Fog images, that’s it’s primary purpose. However - it allows you to create custom boot menu entries. So you could create a boot menu item to do the installation that you’re already doing, probably with very little difficulty. But I have to tell you - I can deploy a Linux image from FOG in under 1 minute. Is the net-deploy faster? Just things to consider. There will also be a FOG Client for linux in the future that can manage names, printers, snapins, joining domains, and other things. @jbob can tell you all about it.
also - more information: