I am trying to deploy a cluster with MAAS.
I installed MAAS on my master node, and the nodes I want to commission boot from PXE.
The master node has two NICs which are set up as follows:
- fabric-0 with 192.168.42.0/24
- fabric-1 with 192.168.43.0/24
The 192.168.43.0/24 is connected to my work notebook which runs a Squid proxy.
My work notebook has 192.168.43.254 on the respective interface, the MAAS master 192.168.43.100.
I now set up a DHCP range on the fabric-0 network, and this is the NIC my nodes I want to install are connected to.
I set Proxy to Peer, and entered
When I now boot a node on the 192.168.42.0/24 network, it boots fine and begins to download packages in the initial steps. But when I command to deploy a system to the node, when the deploying starts, the node tries to use
192.168.43.100 as a proxy - i.e., from fabric-1, which fails as it isn't connected to this network.
Why is that so? Is that a bug or am I missing something here?
Should I manually install and configure a Squid on the MAAS master to listen on the 192.168.42.0/24, use the correct cache_peer and use this as an "external" proxy?