v18: Bug with data counting in firewall rules?

Hi,

I am noticing a strange behavior in v18 and the data counting in the firewall rules. I have some incoming rules (from Internet to DMZ) that are coupled with corresponding DNAT rules. The DMZ contains webservers, so they send a lot more data than they receive. However, the counters in the rules are the other way around: They show a lot more incoming data than outgoing data. 

Unless I am completely misinterpreting these counters (which I would like to rule out), it appears to me these counters have been reversed, e.g. incoming is actually showing outgoing, and outgoing is showing incoming. 

Any thoughts?

  • Hi cryptochrome, 

    Could you please share the screenshot of the firewall rule that shows data in and out counters? I will verify if it should be like that or not and update you.

    Thanks,

  • In reply to H_Patel:

    H_Patel

    Hi cryptochrome, 

    Could you please share the screenshot of the firewall rule that shows data in and out counters? I will verify if it should be like that or not and update you.

    Thanks,

     

    Sure, here you go:

    Firewall Rule:

    Corresponding NAT rule:

    Firewall Rule Details:

     

    Note that this is just one example. I am seeing the same "reversed" counter on other incoming rules. 

    How do you count the data? If someone on the internet initiates the connection and transfers a lot of data, does that count as incoming or outgoing?

  • In reply to cryptochrome:

    Hi  

    To answer your question about how the data counter works, if someone on the internet uploads to your webserver, that would be incoming traffic.  The server responses to those connections will be less unless you are downloading data from internet to local LAN.

    I hope that clears it up.

     

    Thanks!

  • In reply to KingChris:

    My Plex rule (and others) is the same, and I have always been so confused.  When I was on v17, I noticed it was only the "Business" rules which did this.  The other rule type (User / Network??) was the other way around, which is what I would expect.  So it has to do with DNAT rules it seems.  So confused.

    Just now I checked and it shows "in 42GB, out 371MB".  My Plex server itself is not taking in that much data, but serving up to family/friends.  Why is this counted as "in"?  I can't wrap my head around it.  This is clearly 42GB leaving my Plex server, hitting the XG and then going out the WAN.

  • In reply to NateP2:

    *Edit* To reflect this Thread better;

     

     

    V17.5 Client to Server through DNAT.

    Upload Something:

    Download something:

    Same behavior like V18. 

     

     

    LAN to WAN. Download shown as IN Bytes.

    DMZ to WAN. Download shown as IN Bytes.

    WAN to DMZ(DNAT). Download shown as IN Bytes. 

     

    The Plex Scenario shows this perfectly. 

    If the Client is in LAN, Plex in DMZ, streaming a Movie shows the movie as IN Bytes. 

    If the Client is in WAN, Plex in DMZ, Streaming a Movie shows the movie as IN Bytes. 

     

    Just to be complete.

    Client - XG1 SNAT - Internet - XG2 DNAT - Server 

    Client Upload:

    XG1 (SNAT):

    XG2 (DNAT):

     

     

    Client Download something:

    XG1 (SNAT):

    XG2 (DNAT):

     

  • In reply to KingChris:

    KingChris

    Hi  

    To answer your question about how the data counter works, if someone on the internet uploads to your webserver, that would be incoming traffic.  The server responses to those connections will be less unless you are downloading data from internet to local LAN.

     

     

    If that is the case then my observation is correct and we are looking at a bug here. The data is counted in the wrong directions for connections that are subject to DNAT. See Nate's response, too. He pretty much confirms it. 

  • In reply to cryptochrome:

    To add to my answer above something: 

     

    Actually you could resolve this by a little Switch in a Firewall to give the Administrator the possibility to "Switch" IN / OUT. 

    A little flag to tell the Firewall "This Rule is for NAT!". 

    But this would be just a "easy way out". I would actually hope, this is resolved in a deeper level in the upcoming releases!  :) 

  • In reply to LuCar Toni:

    LuCar Toni

     

    Lets recap the IN and OUT in Firewall Rule quickly.

    IN means, traffic coming from the Destination going through the XG to the Source.

    OUT means traffic, coming from the Source going through the XG to the Destination.

    [...] 

    Because basically the Firewall Rules was used for Traffic behind the XG through XG to the Internet, for example, this made perfectly sense. 

    Your Client is downloading something in the Internet. 

    LAN to WAN: IN 100 GB, Out 1 MB - Because the Internet is sending you traffic and we counting this as IN traffic. 

      

    I am not sure I can follow.... Let me re-iterate what I am seeing:

    Internet <-> XG Firewall <-> DMZ Server

    Internet user downloads a lot of data from the server in the DMZ. XG firewall shows this traffic as INcoming on the corresponding firewall rule. Nothing about that makes sense, it's just utterly wrong. 

    Again, this only applies on rules that are subject to DNAT. On SNAT connections, it shows the correct directions. This is inconsistent behavior at best, but I would actually go as far as saying this is clearly a bug.

    Note that this is reflected in all reports and statistics (counters in diagnostics etc.) as well. 

  • In reply to cryptochrome:

    Your Example is the same issue like the the PLEX issue here in this Thread.

     

    Firewall makes perfectly sense for "none DNAT traffic" because the counter was build for everything behind XG. 

    DNAT turns the switch and allows traffic coming from WAN to LAN. 

     

     

    Think about this: you create a ANY - ANY Rule. 

    This will allow LAN to WAN. But also WAN to LAN (Your DNAT Traffic). 

    How should the traffic counter now work? It is hard to tell and that is causing this issue for now. 

     

    Firewall rule counter was not designed to reflect such kind of traffic in the first place. 

    PS: I am not saying it is good in any way. I just try to tell you the Root Cause of this. 

  • In reply to LuCar Toni:

    LuCar Toni

     

    Think about this: you create a ANY - ANY Rule. 

    This will allow LAN to WAN. But also WAN to LAN (Your DNAT Traffic). 

    How should the traffic counter now work? It is hard to tell and that is causing this issue for now. 

     

    How the traffic should be counted? Like on any other network device on this planet, like on any other firewall: based on traffic direction, which can be observed by looking at the interfaces on which traffic enters and leaves. Even simple layer 2 switches can distinguish between incoming and outgoing traffic. There is no guess work involved. You look at the traffic, you see that the connection was initiated from the internet (wan interface) and that reply packets are coming from a different interface. In a nutshell, this is very basic connection tracking. It has nothing to do with NAT. You can clearly tell which side initiates the connection and which side responds. Whether IP addresses get translated in the process is irrelevant. 

    All the firewall has to do is look at who initiated the connection. If the server in the DMZ initiates a connection through an interface, it's outgoing traffic. If the server responds to a connection, it's incoming traffic. 

  • In reply to cryptochrome:

    I fully agree with . Lucar yours response is not helpful!

    If this is an issue, must be addressed and fixed. I guess that the counters are written inside postgres so I do not want to imagine that even the reports are wrong.

    Regards

  • In reply to cryptochrome:

    All the firewall has to do is look at who initiated the connection. If the server in the DMZ initiates a connection through an interface, it's outgoing traffic. If the server responds to a connection, it's incoming traffic. 

     
    That is how XG performs the Traffic logging.  
    NAT is not the issue, it is the definition of Zone concept. The Firewall with IN/OUT as "Terms" causing this misleading terms. 
     

    We are counting the Traffic after processing the Traffic through all modules. It is about to Leave the destination Interface and gets tracked.

     

    Important is the Traffic Initiator, as you said. 

     

     

    LAN to WAN. IN means Download (Send by the XG LAN Interface / WAN Server). OUT means Upload (Send by the XG WAN Interface / LAN Client). 

    WAN to DMZ. IN means Download (Send by the XG WAN Interface / DMZ Server). OUT means Upload (Send by the XG DMZ Interface / WAN Client) 

     

     

    LAN to WAN:

    Download Scenario: Client sends 5 Bytes to XG. Server sends 10 MB to XG.  XG Sends 5 Bytes to Server, XG Sends 10 MB to the Client. 

    From XG Perspective: 

    Firewall: IN 10 MB, OUT 5 Bytes. 

    Firewall: Sent 10 MB to the Client, Sent 5 Bytes to the Server. 

    Firewall: Gets 10 MB from the Server, Gets 5 Bytes from the Client.

    Firewall: Downloads 10 MB from the Server, Uploads 5 Bytes to the Server.

    Firewall: Downloads 5 Bytes from the Client, Uploads 10 MB to the Server.  

     

     

     

    DNAT Scenario:

    Download Scenario: WebClient sends 5 Bytes to XG. DMZ Server sends 10 MB to XG.  

    Firewall: IN 10 MB, OUT, 5 Bytes. 

    Firewall: Sent 10 MB to the WebClient, Sent 5 Bytes to the DMZ Server. 

    Firewall: Gets 10 MB from the DMZ Server, Gets 5 Bytes from the WebClient.

    Firewall: Downloads 10 MB from the DMZ Server, Uploads 5 Bytes to the  DMZ Server.

    Firewall: Downloads 5 Bytes from the WebClient, Uploads 10 MB to the WebClient.  

     

     

    It is more likely a perspective issue. 

     

    Looking at the Plex Issue.

    Plex Client is in the WAN. 

    Plex Server is in the DMZ. 

     

    Client is building the Connection to Plex Server.

    Starting to Stream a Movie. 

    Client requests the Movie, sending 5 MB. 

    Server sending the Movie to the Client, Sending 10 GB. 

    Firewall: IN 10 GB, OUT, 5 MB. 

    Firewall: Sent 10 GB to the Plex Client, Sent 5 MB to the Plex Server. 

    Firewall: Gets 10 GB from the Plex Server, Gets 5 MB from the Plex Client.

    Firewall: Downloads 10 GB from the Plex Server, Uploads 5 MB to the  Plex Server.

    Firewall: Downloads 5 MB from the Plex Client, Uploads 10 GB to the Plex Client.  

     

     

     

    XG Firewall Tracking does everything correct, from my point of view. 

     

     

    Maybe remove IN/OUT as Terms? You would have: 10 GB / 5 MB for example. 

     

     

     

    *edit* 

    Lets take another example:

    Client to DMZ. 

    Client downloads a file from my DMZ Server.

    IN would be the Traffic send by my Server.

    OUT would be the traffic send by my Client. 

     

    Thats how it works since the beginning. 

    Now the same is done by DNAT + Firewall. 

    Basically XG did not Change anything in his firewall behavior for NAT. It is simply another Client to Server concept. 

  • In reply to LuCar Toni:

    LuCar Toni

    It is more likely a perspective issue. 

     

     
    Maybe. And just to be clear: I completely understand what you are saying, but I think your perspective is still wrong. It's overly complicated, illiogical, entirely disregards the concept of TCP/IP session initiation (from which you must derive traffic direction) and goes against what any other firewall vendor is doing. As a matter of fact, it goes against what every vendor of any kind of network device is doing. 
     
    Just to make sure, it must not matter whether a connection is subject to NAT. All that matters is who is initiating the connection. From that perspective, session initiator is always IN, and session responder is always OUT. 
     
    To stay with your example on Plex:
     
    Plex Client initiates connection to Plex Server (sends 50 KB of data, requesting playback of movie). Plex Server ACKnowledges session, responds with 10 GB of data. Firewall is stateful, sees which side is initiating (SYN) and which side is responding (SYN ACK). Hence, Plex Client is INcoming traffic, Plex Server is OUTgoing traffic. 
     
    If the Plex Server would terminate the client session and then establish a new session back to the client to send the movie, your logic would be correct. But that's not how it works. 
     
    To be frank, I don't understand why we even have to discuss this, given the fact that XG does it correctly for all connections, except when NAT is in place (which changes absolutely nothing in terms of session establishment). This has to be looked at from a network perspective and a user perspective, not from a developer's perspective. 
     
  • In reply to cryptochrome:

    Plex Client initiates connection to Plex Server (sends 50 KB of data, requesting playback of movie). Plex Server ACKnowledges session, responds with 10 GB of data. Firewall is stateful, sees which side is initiating (SYN) and which side is responding (SYN ACK). Hence, Plex Client is INcoming traffic, Plex Server is OUTgoing traffic. 
     
    XG uses the Client Perspective. Hence it uses the terms of IN and OUT from the Client perspective. (Initiator). 
    Plex Client traffic is OUTgoing Traffic, because the initiator sends this traffic.
    Plex Server traffic is INcoming Traffic, because the recipient sends this traffic. 
     
    That is how XG does it for years and nobody was asking to change it. 
     
    Now in the setup of the DNAT, it could be misleading. DNAT gives us the setup, that the Initiator is NOT behind XG, instead the recipient is behind XG. 
     
     
    I am actually not sure, how V17.5 (Business App Rules) used to work. 
  • In reply to LuCar Toni:

    LuCar Toni
     
    XG uses the Client Perspective. Hence it uses the terms of IN and OUT from the Client perspective. (Initiator). 
     

     
    Well that perspective is wrong. Especially given the fact that you use this perspective only on DNAT traffic, but not on SNAT traffic (where you actually do it correctly). Totally inconsistent. 
     
    LuCar Toni
     
    Plex Client traffic is OUTgoing Traffic, because the initiator sends this traffic.
    Plex Server traffic is INcoming Traffic, because the recipient sends this traffic. 
     
    That is how XG does it for years and nobody was asking to change it. 
     
     
    As you can see, I am not the only one who is confused about this weird perspective of looking at things. This is probably coming from a developer's mindset, not one of a network engineer's. 
     
    LuCar Toni
     
    Now in the setup of the DNAT, it could be misleading. DNAT gives us the setup, that the Initiator is NOT behind XG, instead the recipient is behind XG. 
     
     
    Right. And this should be changed. If DNAT is "misleading" the firewall's data collection, then it's a bug. 
     
    Just to re-iterate, when there is no DNAT in place, the XG actually does it right. The traffic directions are correct in all cases except when DNAT is applied.