Hi @davosian1
Thanks again for taking on this project and being so diligent with it.
I suspect what is happening is that your default network is getting selected, and in your case that just happens to be the one you aren’t interested in. Here are a couple of settings I think you could use to force the network selection you want. I’d start with network_interface. I found this relevant doc snippet in the upgrade guide that I think explains why.
“If you have several routable IPs, it is advised to configure the client’s network interface such that tasks bind to the correct address.”.
- network interface - This specifies the name of the interface to force network fingerprinting on for the whole client. When run in dev mode, this defaults to the loopback interface. When not in dev mode, the interface attached to the default route is used. The scheduler chooses from these fingerprinted IP addresses when allocating ports for tasks. This value support go-sockaddr/template format.
So in your case it would be something like
client {
network_interface = "ens10"
}
- host_network - This is used to register additional host networks with the node that can be used when port mapping. Note that there is a corresponding field on the jobspec that I think will also have to be set to hook into this network. If you need/want to bind to multiple host networks, I think this is the way you can achieve that.
Did this get you where you needed to go?