Subnet/VLANs with managed and unmanaged switches
-
@ g: It sounds like you are not really sure how to setup your switch. I would reset it to factory defaults and start simple. Maybe you somehow have enabled 802.1x without setting up a radius server for this and now the ports are locked down.
@ cirrusflyer: This is bad network design and you usually should not do that. However it will be available in the upcoming version of pfSense 1.3. If you have such a setup now you could "fake" it by using 2 nics (or vlans) that hook up to the same switch/layer2 network unless you want to fiddle around with console commands which will be gone after reboot or filter changes and so on…
-
@ g: It sounds like you are not really sure how to setup your switch. I would reset it to factory defaults and start simple. Maybe you somehow have enabled 802.1x without setting up a radius server for this and now the ports are locked down.
Hoba, yes I am not sure about somethings. It is still set to factory defaults. When I tried to change a port from default VLAN1 to VLAN100 untagged, it complained about an unauthenticated/untagged port. So I thought maybe some authentication method was necessary.
-
Got it all working. Thanks a lot for the help.
What I was hoping for was a seamless transition between the old and new router. Since the LAN interface only contains VLANs, accessing the rest of the network goes thru WAN. So it's definitely not possible to have the pfSense LAN be re0 AND have VLANs attached to re0? I can't get traffic thru to pfSense that way. Either way, does the switch port need to be trunked or can it be set to general with default VLAN1 untagged and the rest tagged? I sort of undertand why it wouldn't work but not exactly.
If I wanted to achieve all of this, it sounds like I could use 3 NICs. WAN, LAN, and the all the VLANs on the 3rd. And then a 4th for CARP. That's the only way?
-
only the port connected to pfsense needs to be tagged
-
So it's definitely not possible to have the pfSense LAN be re0 AND have VLANs attached to re0?
It is possible. But you shouldnt do it.
But you can define your LAN as a VLAN.
This way you would need only 3 Interfaces:
-WAN
-Interface_with_all_VLAN_inclusive_LAN
-CARP_sync_interfaceTo make it more clear:
|–------WAN----------------|
| | |
| pfSense --- CARP_sync
| | |
|------VLAN_interface-------|
|
|
trunk (tagged)
|
|-----|----------|
| VLAN_switch |
|--|----|-----|--|
/ |
/ | (untagged)
(untagged) |
/ (untagged)
LAN | OPT2
OPT1You have NO untagged traffic on the VLAN_interface. You just dont assign the interface directly. Only VLAN's which are on this interface.
-
So it's definitely not possible to have the pfSense LAN be re0 AND have VLANs attached to re0?
It is possible. But you shouldnt do it.
I think I tried the right way to do it but it didn't work.
LAN - re0
VLANs attached to re0
switch port trunkedNo traffic gets to the router. I think I also tried general (full 802.1q) with all VLANs on the switch.
I understand not having a non-VLAN interface is more secure. Any other reason?
But you can define your LAN as a VLAN.
I think your original suggestion was to define LAN as a VLAN? That was done and everything works. The problem we're migrating and so many networks will be VLAN unaware until we're finished. Let's say:
192.168.0.x - server network
192.168.10.x
192.168.20.x
192.168.30.xLet's say I move 10 to VLAN10. If I want to route to any other network, since the LAN is a VLAN, it routes to WAN to access any other network. I'm not sure if untagged traffic on the 0 network reaches the interface. I'm not sure how to deal with that. Another NIC would work for sure. But when I try:
pfSense:
LAN - 192.168.0.254 - re0 real interfacere0:
VLAN10, 20, 30, …From the switch, I can't reach any of the networks.
The main goal is migration without switching everything at once.
-
IMO you should migrate it all together since the firewall will be the backbone. So setup a lab environment where you can test your setup.
-
There shouldn't be any difference in running the lan assigned to a real nic or a vlan nic.
The maximum amount of traffic to push between subnets is dependent on this. Don't know his needs so maybe a shared link simply is not enough.
-
192.168.0.x - server network
192.168.10.x
192.168.20.x
192.168.30.xLet's say I move 10 to VLAN10. If I want to route to any other network, since the LAN is a VLAN, it routes to WAN to access any other network. I'm not sure if untagged traffic on the 0 network reaches the interface. I'm not sure how to deal with that. Another NIC would work for sure. But when I try:
pfSense:
LAN - 192.168.0.254 - re0 real interfacere0:
VLAN10, 20, 30, …From the switch, I can't reach any of the networks.
It wont route to WAN.
pfSense just routes between it's interfaces directly. And each VLAN is treated as an interface.What do you mean from the switch you cannot reach the other networks?
Did you set the default gateway to pfSense?
And created rules on the (VLAN)interface that allow traffic?I have this exact setup working.
I'll post screenshots of the config of my switch when i get home.The maximum amount of traffic to push between subnets is dependent on this. Don't know his needs so maybe a shared link simply is not enough.
There are a lot of VLAN capable switches that offer 2 or 4 Gbit port.
I use such a setup where the trunk interface is Gbit and the normal ports on the switch are 100 Mbit.
You can avoid this bottleneck. -
Temporarily. the WAN interface was on 192.168.0.x. So it reaches that network. But it's not the best way.
With the setup you gave above:
re0 LAN - VLAN10
re0 multiple VLANs attachedI still need to be able to read the 0.x network. Those ports are connected as default VLAN1 on the switch. I also tried making VLAN1 re0 LAN; that didn't work. Maybe I was missing something. Maybe pfSense was properly setup but the switch wasn't. It seemed tagged VLAN traffic went thru but default VLAN1 untagged didn't.
-
Those ports are connected as default VLAN1 on the switch.
They are not supposed to be the default VLAN1.
VLAN1 (default) are all the ports that are "not in a VLAN".
But the point of moving the LAN to a VLAN is: NOT USING VLAN1
VLAN1 is a reserved VLAN!
Refer to the 802.1Q specs page 76.
http://standards.ieee.org/getieee802/download/802.1Q-2005.pdfTable 9-2—Reserved VID values
VID value(hexadecimal) Meaning/Use
0 The null VLAN ID. Indicates that the tag header contains only priority
information; no VLAN identifier is present in the frame. This VID value shall not
be configured as a PVID or a member of a VID Set, or configured in any Filtering
Database entry, or used in any Management operation.1 The default PVID value used for classifying frames on ingress through a Bridge
Port. The PVID value of a Port can be changed by management.FFF Reserved for implementation use. This VID value shall not be configured as a
PVID or a member of a VID Set, or transmitted in a tag header. This VID value
may be used to indicate a wildcard match for the VID in management operations
or Filtering Database entries.You wrote that you've set LAN to VLAN10
–> Just set the PVID for all the ports that should be LAN to 10.Then tag packets going to pfSense (on the trunk) and untag packets going to the clients.
-
Those ports are connected as default VLAN1 on the switch.
They are not supposed to be the default VLAN1.
VLAN1 (default) are all the ports that are "not in a VLAN".
…
But the point of moving the LAN to a VLAN is: NOT USING VLAN1:) yes exactly. But since I'm not making the complete switch yet (moving one network at a time), I still need to access VLAN1, everything that is not on a VLAN. So what I was hoping to do is have one interface on pfSense that would have access to everything not on a VLAN and all the VLANs. :)
-
Aha now i get it ;D
Well i suppose as long as it's only temporarily you could assign the interface directly.
Of course a second interface would work too.It's not like that it wont work, it's just "bad" design.
"bad" as in mixing tagged and untagged traffic on the same wire.But i dont know if you want to add a 4th NIC just to make the transition :)
When i thinking about it: can you set the trunk on your switch that it eggresses tagged VLAN1 packets?
Or do you mean withI still need to be able to read the 0.x network. Those ports are connected as default VLAN1 on the switch. I also tried making VLAN1 re0 LAN; that didn't work. Maybe I was missing something. Maybe pfSense was properly setup but the switch wasn't. It seemed tagged VLAN traffic went thru but default VLAN1 untagged didn't.
exactly that?
Because it should be possible from the pfSense side to accept tagged VLAN1 packets.
-
Aha now i get it ;D
..
But i dont know if you want to add a 4th NIC just to make the transition :)I may just do that.
When i thinking about it: can you set the trunk on your switch that it eggresses tagged VLAN1 packets?
Or do you mean withI still need to be able to read the 0.x network. Those ports are connected as default VLAN1 on the switch. I also tried making VLAN1 re0 LAN; that didn't work. Maybe I was missing something. Maybe pfSense was properly setup but the switch wasn't. It seemed tagged VLAN traffic went thru but default VLAN1 untagged didn't.
exactly that?
Because it should be possible from the pfSense side to accept tagged VLAN1 packets.
I have to try again to see if the Dell switch can tag VLAN1 traffic to the port. I tried setting the Dell to tag traffic to a particular port but it seemed unable to. I tried setting the switch port to something besides trunk (general I think with all ports going tagged) and I think that didn't work.
I may try again with VLAN1 on the interface. I was curious about a earlier comment about not having interfaces directly assigned.
Should the LAN interface be assigned anything or can just all the VLANs be attached to interface re0? and LAN have nothing?
-
I may try again with VLAN1 on the interface. I was curious about a earlier comment about not having interfaces directly assigned.
Should the LAN interface be assigned anything or can just all the VLANs be attached to interface re0? and LAN have nothing?
That goes into the same as
It's not like that it wont work, it's just "bad" design.
"bad" as in mixing tagged and untagged traffic on the same wire.You "should" not assign an interface on which VLANs are running.
Like i said: It will work. It's just not good network design.Either Have LAN as VLAN too,
or have another interface as LAN -
Either Have LAN as VLAN too,
or have another interface as LANBy LAN you mean the the LAN itself and not the pfSense LAN interface?
I can go with an additional NIC until we fully switch over but I'm still curious.
I haven't looked into CARP yet. It seems that I would have a CARP ip for each VLAN and WAN and then use that as the default gateway for clients?
I'm almost there. :) Really can't wait to start using pfSense.
-
Lets refer to the physical interface as re0.
I mean: asign the logical LAN-interface either as VLAN on re0, or add another NIC (re1) and assign the LAN-interface directly to re1.CARP is not what you are looking for.
CARP is used for redundant hardware. (Failover on hardware-fail)
Or to create Virtual IPs to/from which you NAT stuff.Each VLAN is a seperate logical interface on pfSense.
Meaning each interface will have its own IP. -
Lets refer to the physical interface as re0.
I mean: asign the logical LAN-interface either as VLAN on re0, or add another NIC (re1) and assign the LAN-interface directly to re1.That's the plan. Should be able to get around to it later today or tomorrow.
CARP is not what you are looking for.
CARP is used for redundant hardware. (Failover on hardware-fail)
Or to create Virtual IPs to/from which you NAT stuff.I would create a Virtual CARP IP on each VLAN interface and then use that as the default route for each VLAN? The idea would be to avoid routing to any real IPs yes?
-
I would create a Virtual CARP IP on each VLAN interface and then use that as the default route for each VLAN? The idea would be to avoid routing to any real IPs yes?
I dont follow.
What is the point of having a router if you dont want to route? -
I would create a Virtual CARP IP on each VLAN interface and then use that as the default route for each VLAN? The idea would be to avoid routing to any real IPs yes?
I dont follow.
What is the point of having a router if you dont want to route?It's just that I haven't read the docs yet. For failover to another router, I would want the default gateway interface on each network to move between routers? So if 192.168.[VLAN].1 was the default route for each network, how would this failover to the 2nd router? By using CARP IPs attached to each VLAN interface?
