This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Classification of traffic using NTP fails

Hi folks,

I have tried various settings in the firewall rules that use the hairpin NAT to provide local NTP services. The daily reports show a high count of hits on UDP 123. Occasionally the report shows a low count hit on NTP, which I suspect is from the internal NTP server checking time.

The issue is not new and has been seen in a number of previous versions of XG firmware.

So, what needs to be changed in my rules to allow correct classification of NTP or is there a fix required to XG firmware?

The current hairpin rules are using DPI, I have tried using web proxy.

Ian



This thread was automatically locked due to age.
  • Correct classification or even classification is not important?

    ian

    XG115W - v20 GA - Home

    XG on VM 8 - v20 GA

    If a post solves your question please use the 'Verify Answer' button.

  • Most likely the app classification is not accurate. But mine is? Maybe your traffic is different to mine. NTP is not NTP in most cases. There are different types of NTP. So based on the port, plenty attackers try to use the port 123 to communicate with there apps (C2 Communication). 

    If your NTP is not correctly pickup, you could do a wireshark dump and check the wireshark dump, if you find anything odd looking. 

    There were some issues with ThunderVPN etc. based on Port123. But not picking up NTP at all is rather new to me. 

    Check the logviewer on Port123 if there is a app classification. 

    __________________________________________________________________________________________________________________

  • I checked the logviewer and found only one device uses Network Time Protocol and that is a power controller. All other devices are not recognised as using NTP but port 123 and this includes the two Sophos APX120s, the NTP server running a specific NTP server function.

    Thunder VPN has not surfaced for a couple of releases.

    I did find a couple of errors in the firewall rueles which have been cleaned up. The APX that is managed by CM continually does NTP checks where as the locally managed APX does not.

    Ian

    XG115W - v20 GA - Home

    XG on VM 8 - v20 GA

    If a post solves your question please use the 'Verify Answer' button.

  • That is expected. NTP is used for time correction. As Central is using a lot of certificates, you need to fetch the time first, otherwise you could run in one of the oldest TLS issues (time of certificate is not valid). 

    Do a tcpdump of this traffic and in another shell do a conntrack. Then we could take this to Labs to analyse the pattern. 

    __________________________________________________________________________________________________________________

  • I have been trying to capture the traffic using PCAP, but all attempts fail with no records found, so in summary I must be doing something wrong, but not obvious to me.

    I will try tcpdump later today.

    Ian

    XG115W - v20 GA - Home

    XG on VM 8 - v20 GA

    If a post solves your question please use the 'Verify Answer' button.

  • I have captured some PCAP traffic and contract but not of the same connections.

    Looking at the PCAP results would suggest that the XG has an issue because of the data included or omitted from the capture.

    Ian

    XG115W - v20 GA - Home

    XG on VM 8 - v20 GA

    If a post solves your question please use the 'Verify Answer' button.

  • I have sent you a PM with the conntrack file.

    Ian

    XG115W - v20 GA - Home

    XG on VM 8 - v20 GA

    If a post solves your question please use the 'Verify Answer' button.

  • We need a pcap file. Not a screenshot of some packet capture. 

    support.sophos.com/.../KB-000037007

    __________________________________________________________________________________________________________________

  • I have added a tcpdump file in a PM to you.

    I think I can maybe see the cause, the traffic is decoded as NTPv4, not NTP, so maybe a there is a need to add a couple of new classifications of NTPv4 

    Ian

    More analysis, shows the device that records the traffic as Network Time Protocol in logviewer is using NTPv3.

    XG115W - v20 GA - Home

    XG on VM 8 - v20 GA

    If a post solves your question please use the 'Verify Answer' button.

  • I have sent you a tcpdump of the IP4 traffic in a PM.

    I have attached a screenshot of the tcpdump capturing NTP using IPv6.

    Ian

    XG115W - v20 GA - Home

    XG on VM 8 - v20 GA

    If a post solves your question please use the 'Verify Answer' button.