Hi,
I have configured several Subnets on my RB3011. All Subnets cannot see each other, it is disabled by FW-Rule.
Now I would like to configure some exceptions. I have a local SIP Server on Subnet1 with IP: 192.168.1.10. Client on Subnet1 can connect correctly to the Server, but Clients on Subnet2(192.168.2.0/24) do not work. This works as expected.
How do I have to configure a common rule for Subent2 to get this working? I think I have to route UDP-Port 5060.
Can some help here?
Christian
It would be useful to see/post your config to understand how your subnets are created.
I am assuming you are saying they are blocked by FW rules that you mean the last rule in your Foward Chain is drop all else rule??
but make sure there’s a rule accepting connection-state=established,related before the new rule. And move the new rule above the rule which currently prevents connectivity to happen.
@anav showed a rule which logs all conections triggering this rule. While it’s fine to debug it shouldn’t be set up for production … it will litter the log with irrelevant entries making you miss the important ones.
Good point mkx, yes the log prefix enabled was simply to verify if the forwarding is taking place as desired. Once up and running it should be removed.
As also demonstrated by MKXs rule, the originating source can be identified in a few ways.
1 - source address (an individual address or whole subnet)
2 - in-interface (the LAN where the request is coming from/originating)
(Ive used a combo of both in the past for better security but have been told its simply wasteful and thus the source address as shown is adequate).
3. Use of protocol and ports is a good idea in terms of limiting the ports and protocols allowed by the originating LAn to hit your server.
(seeing as there is no other functionality of the server, I tend to think that this simply wasteful as well unless someone can point out why otherwise ???
Performance-wise it might indeed be wasteful. However, you never know if in future that same server might get some other functionality you don’t want to expose to the said subnet. This case, however, is not similar to yours (filtering based on both physical port and IP address): in “my” case filter is only very specific (narrowing down possibilities) - but in this very particular case it may not be necessary to be that very specific. In “your” case, both filter criteria (physical port and IP address) are, strictly speaking, not related at all. Specifying both might narrow down the possibilities (possibly breaking L2 network in unlucky case) or it may broaden possibilities (in another unlucky case).
Hi,
many thanks for your replies and sorry for late response, but I am on business trip and limited access to internet and my router. I will check tomorrow and post my config.
Thanks for help,
Christian
No worries spartacus, in the meantime I am simply sparring with mkx over some minutia.
So mkx,
The questions comes down to what is most efficient.
a. asking the CPU to inspect the forwarded packet for IP, protocol and port (and discarding those that do not fit) OR
b. asking the CPu to inspect the forwarded packet for IP only (and discarding those that do not fit).
The questions I have is why would anyone on the originating LAN be hitting that IP address with anything other than the right protocol and port?
Does the admin expect some folks on the originating LAN to try and hit the server with different types of traffic (hack in?) ?
c. What is more efficient wrt to CPU load?
To have a rule that inspects a packet for allowable IP (which it does) but it has wrong port or protocol and drops it?
VERSUS
The packet is passed to the server (allowable IP) but the packets hits a closed port, or even an open port but with the wrong protocol.
I believe the latter events are transparent to the CPU (no load).
@anav, I don’t see continuing the discussion being productive. The baseline is: how paranoid should network administrator be to keep networks safe and going versus available (and allowable) resources?
My stand is that administrator should be as paranoid as possible and that security of a network should not be based neither on users’ friendliness nor assumptions about fair use (most of known exploits target such pitfalls).
Ahh well okay, I sense you are raising the white flag on efficiency and diverting to the ole security trumps all standby.
Perhaps we can do combat another day.
Hi everybody,
thanks for support. But it seems to be not working. The PC Client Software, Phoner, is working on the Client Subnet (192.168.2.0/24) and the Server IP is 192.168.1.10. I configured full access betwee the subnets without success:
/ip firewall filter add action=accept chain=forward dst-address-list=VlanFriends in-interface-list=LAN src-address-list=VlanFriends comment="Allow inter VLAN communication with VLAN friends"
Both networks are participants from VlanFriends.
Any ideas?
Christian
Hi I am very much interested in this rule…
add action=accept chain=input comment=“Allow LAN NTP queries” dst-port=123
in-interface-list=LAN protocol=udp
What is its purpose? I have a smart device (honeywell leak detector) that refuses to connect to honeywell via my CapAC wifi and hex router.
I do it through a consumer cisco linksys router with no problems. Some ubiquiti forums talk about NTP and port 123 but I dont block that at all in any forward rule.
Thus it intrigued me to see you have this as a lan to router (input rule). Perhaps I could try this to see if it works for my smart device???
+++++++++++++++++++++++++++++++++++++++++++++++++++++
Back to the show!! The first thing is that your vlan config information is missing so that is not helpful
This rule allows any queries coming from/originating on vlan30 (which has no structure (no config viewable) to (??? on the specific ports. Without the VLAN config I am lost here…
the NTP rule is necessary to accept NTP queries accross my subnets
I do not use Port 25 for Email
Port forwarding rule is disabled
The SONOS Rules are only for my Sonos devices in VLAN30 (192.168.30.0). It is necessary for the UPNP Discovery and forwards the requests to and from the other Subnets. This works and has nothing to do with the SIP protocol. No SIP Client is in VLAN30.
But your nTP rule is on the input chain??
If your devices need to query NTP dont they simply go out to the internet via the forward chain?? (in other words part of normal traffic and thus no rule required??)?
Okay so if NTP query is done on UDP, is that covered in forward chain rules? Does mikrotik recognize a device going on on UDP123 as a new connection and thus will be tracked??
Hi,
maybe there is a missunderstanding for the NTP-Rule:
NTP-Client (to sync RB with NTP-Servers in WAN) and NTP Server (for Clients in LAN) are running on RB3011 and all clients accross the different subnets can use the GW-Address of the current Subnet als NTP-Server to synchronize the time. Therefore I need the Input Rule on RB. It works!