Hi, first want to say I'm very impressed with Proxmox and looking to migrate away from Hyper-V to open source alternatives. I have an existing network set up already and trying to understand how to fit Proxmox into it as I'm having trouble getting VMs connected.
I have multiple networks and multiple NICs in the PVE host. I would have this on my old Hyper-V host as well, and would define one network card as the "host's" NIC, and the other NICs would be linked to virtual switches to which VMs would attach. The option for the host to share these virtual switches was disabled. Each virtual switch was effectively a dumb switch connected to a VLAN'd port on a physical switch.
That is where I have come from, and trying to recreate the equivalent in PVE. I hope this doesn't come across as a "BUT THIS IS HOW MICROSOFT DOES IT" post, if it doesn't compare I'm here to learn. If it's done differently to Hyper-V that's what I'm leaving behind and hope to understand the PVE way.
Per the documentation I've created a bridge on the "main" NIC, so it's sharing with the host, and defined a network range with the CIDR and gateway fields to match the physical network. This didn't work (no communication to the network, no DHCP assignment from the existing network, static IP makes no difference) so I tried creating a bridge on a second NIC where I left blank the CIDR and gateway blank as I would have done with the old host, in Hyper-V the minimum configuration of creating a virtual switch and attaching a physical NIC were all that was required for a VM on the virtual switch to communicate.
So is this possible? The documentation has examples of /etc/network/interfaces with vmbrX assigned IP addresses etc. Can I have one NIC dedicated to the host, and other NICs acting as dumb switches to VMs with no awareness of these NICs at IP level?
Thanks in advance, any help appreciated.
I have multiple networks and multiple NICs in the PVE host. I would have this on my old Hyper-V host as well, and would define one network card as the "host's" NIC, and the other NICs would be linked to virtual switches to which VMs would attach. The option for the host to share these virtual switches was disabled. Each virtual switch was effectively a dumb switch connected to a VLAN'd port on a physical switch.
That is where I have come from, and trying to recreate the equivalent in PVE. I hope this doesn't come across as a "BUT THIS IS HOW MICROSOFT DOES IT" post, if it doesn't compare I'm here to learn. If it's done differently to Hyper-V that's what I'm leaving behind and hope to understand the PVE way.
Per the documentation I've created a bridge on the "main" NIC, so it's sharing with the host, and defined a network range with the CIDR and gateway fields to match the physical network. This didn't work (no communication to the network, no DHCP assignment from the existing network, static IP makes no difference) so I tried creating a bridge on a second NIC where I left blank the CIDR and gateway blank as I would have done with the old host, in Hyper-V the minimum configuration of creating a virtual switch and attaching a physical NIC were all that was required for a VM on the virtual switch to communicate.
So is this possible? The documentation has examples of /etc/network/interfaces with vmbrX assigned IP addresses etc. Can I have one NIC dedicated to the host, and other NICs acting as dumb switches to VMs with no awareness of these NICs at IP level?
Thanks in advance, any help appreciated.