endzyme
Cadet
- Joined
- Dec 27, 2023
- Messages
- 5
I am running TrueNAS-13.0-U6.1.
I have a problem where my VMs associated with bridges which are on a VLAN, no longer receive traffic. My configuration is as follows:
VM(vm1) <=> NIC(e1000) <=> bridge400 <=> vlan400 <=> re0(the host Ethernet card)
&
VM(vm2) <=> NIC(e1000) <=> bridge98 <=> vlan98 <=> re0(the host Ethernet card)
The NICs in the VM used to get their IPs via DHCP and communicate with my router fine.
My other VMs which have their NICs associated directly with re0 work fine and if I add a a NIC to "vm1" and "vm2" which is attached to re0, they also work with DHCP.
I have tried diagnosing with tcpdump on both my router and the truenas machine. When inspecting from the router, I do see frames requesting IPs (DHCP requests and responses) coming into and out of the router.
On the truenas machine I see frames on re0 leaving the box, tagged with the correct vlan, and I also see the router responses to DHCP requests which assign the expected IPs for the boxes in question. When inspecting the vnet interface (which is created when I start up the VMs) I only see the requests and none of the responses.
I should also mention that the bridge98 interface has an ip statically assigned in the same subject for that vlan and can not ping the router. Oddly enough I do see the right eth mac address in the arp table, but for some reason the truenas machine doesn't acknowledge the ping responses.
I've also taken the steps of removing all NICs from the VMs which are not working, and re adding them. I've also, separately, removed the vlans and bridges from the truenas, leaving only re0, rebooted, and then readded everything, including new NICs attached to the newly created bridges. Still it exhibited the same tcpdump behavior.
To be honest I haven't looked at all the '-e' options from tcpdump to verify all the mac addresses make sense. I have an inkling that the problem exists on the truenas server itself, something blocking it from passing frames from the bridge to the vnet.
My next hardware step is to connect the server directly to the router which houses the gateway in the vlan to eliminate any oddities from switches, but I don't think it's that because I see the response frames getting to re0 on the truenas server.
I'm not too sure how bsd works for ether layer 2 routing or how vnets work. I want to confirm how the vnet is configured and I want to setup a more simplified test to see if the OS is behaving normally.
Are there any next steps people could suggest?
I have a problem where my VMs associated with bridges which are on a VLAN, no longer receive traffic. My configuration is as follows:
VM(vm1) <=> NIC(e1000) <=> bridge400 <=> vlan400 <=> re0(the host Ethernet card)
&
VM(vm2) <=> NIC(e1000) <=> bridge98 <=> vlan98 <=> re0(the host Ethernet card)
The NICs in the VM used to get their IPs via DHCP and communicate with my router fine.
My other VMs which have their NICs associated directly with re0 work fine and if I add a a NIC to "vm1" and "vm2" which is attached to re0, they also work with DHCP.
I have tried diagnosing with tcpdump on both my router and the truenas machine. When inspecting from the router, I do see frames requesting IPs (DHCP requests and responses) coming into and out of the router.
On the truenas machine I see frames on re0 leaving the box, tagged with the correct vlan, and I also see the router responses to DHCP requests which assign the expected IPs for the boxes in question. When inspecting the vnet interface (which is created when I start up the VMs) I only see the requests and none of the responses.
I should also mention that the bridge98 interface has an ip statically assigned in the same subject for that vlan and can not ping the router. Oddly enough I do see the right eth mac address in the arp table, but for some reason the truenas machine doesn't acknowledge the ping responses.
I've also taken the steps of removing all NICs from the VMs which are not working, and re adding them. I've also, separately, removed the vlans and bridges from the truenas, leaving only re0, rebooted, and then readded everything, including new NICs attached to the newly created bridges. Still it exhibited the same tcpdump behavior.
To be honest I haven't looked at all the '-e' options from tcpdump to verify all the mac addresses make sense. I have an inkling that the problem exists on the truenas server itself, something blocking it from passing frames from the bridge to the vnet.
My next hardware step is to connect the server directly to the router which houses the gateway in the vlan to eliminate any oddities from switches, but I don't think it's that because I see the response frames getting to re0 on the truenas server.
I'm not too sure how bsd works for ether layer 2 routing or how vnets work. I want to confirm how the vnet is configured and I want to setup a more simplified test to see if the OS is behaving normally.
Are there any next steps people could suggest?