#pxe #netwrokboot #dhcp #warewulf #pxe #netwrokboot #dhcp #warewulf


rodrigoceccatodefreitas@...
 

Hello,

I am trying to get a cluster to boot over PXE using Warewulf. I have added the nodes to the Warewulf database and I used $ tftp get to verify that tftpd is running.
When I boot the compute nodes, they get their IP through the DHCP server that is on another computer but find no "PXE server" to boot from.
Every node can reach each other through the network.

I installed the headnode (CentOS 7) using Cobbler. Cobbler and the DHCP server that attributes IPs are on different computers (not the ones that compose the cluster).

I have, for example, the following IPs for the cluster:
10.68.107.100 (headnode)
10.68.107.103 (worker node)

While the DHCP server is on 10.68.254.9.

I do not understand much of PXE and need help fixing this issue. If I missed some relevant information, I'll be glad to provide it as soon as possible.

Thanks in advance,

Rodrigo.

(some errors I got changing the boot order in the bios)



 

The provisioner (headnode?) should be the system that's handling DHCP... If you put a `next-server` entry for the specific nodes in your main DHCP server to point back to the provisioner, that may work depending on how other things are setup...

Also be sure you've killed any firewall, or opened up all the ports, you'll be needing on the provisioner.

Also... what's the network you're tryign to work with here?  10.68.0.0/16

-J

On Mon, May 11, 2020 at 11:56 AM <rodrigoceccatodefreitas@...> wrote:
Hello,

I am trying to get a cluster to boot over PXE using Warewulf. I have added the nodes to the Warewulf database and I used $ tftp get to verify that tftpd is running.
When I boot the compute nodes, they get their IP through the DHCP server that is on another computer but find no "PXE server" to boot from.
Every node can reach each other through the network.

I installed the headnode (CentOS 7) using Cobbler. Cobbler and the DHCP server that attributes IPs are on different computers (not the ones that compose the cluster).

I have, for example, the following IPs for the cluster:
10.68.107.100 (headnode)
10.68.107.103 (worker node)

While the DHCP server is on 10.68.254.9.

I do not understand much of PXE and need help fixing this issue. If I missed some relevant information, I'll be glad to provide it as soon as possible.

Thanks in advance,

Rodrigo.

(some errors I got changing the boot order in the bios)


_._,_._,_


rodrigoceccatodefreitas@...
 

Hello Jason, thanks for the reply!

Yes, the headnode is the provisioner and has Warewulf installed. The network I am working with is 10.68.107.0/8.

There is no firewall running and all ports are reachable between any two nodes in the cluster;

So, I will set 'next-server': 10.68.107.100 on the DHCP server (not the headnode) and report if it worked :)


RicardoMillan
 

Try to change the bios to legacy . Do not use UEFI.

On Mon, May 11, 2020 at 1:15 PM <rodrigoceccatodefreitas@...> wrote:
Hello Jason, thanks for the reply!

Yes, the headnode is the provisioner and has Warewulf installed. The network I am working with is 10.68.107.0/8.

There is no firewall running and all ports are reachable between any two nodes in the cluster;

So, I will set 'next-server': 10.68.107.100 on the DHCP server (not the headnode) and report if it worked :)

--