PA-3050 detected dns queries to malicious URL but all other anti-malware detected none

[SingChung]

The Palo Alto device is saying that a workstation on the network is querying the DNS server for some malicious URLs for some of the dates and some of the time.

Full scan using McAfee VSE, Microsoft Safety Scanner, Malwarebytes, Spybot, said no malware detected.

Using McAfee Getsusp I upload 40 suspicious and unknown files, analysed in McAfee lab – no malware detected. McAfee put these files under Virus Total, antivirus from 57 different vendors all say ‘clean’.

The log that says detection of malicious URL:

Receive Time

Threat/Content Type

Generate Time

Rule

Application

Virtual System

Source Zone

Destination Zone

Source Port

Destination Port

IP Protocol

Action

URL

Threat/Content Name

Category

Severity

Direction

12/3/2017 13:56

spyware

12/3/2017 13:56

POC Alert All

dns

vsys1

TapZone2

TapZone2

55305

53

udp

alert

P2P-Worm.palevo:brero.balkan-hosting.net(3839431)

any

medium

client-to-server

12/3/2017 13:56

spyware

12/3/2017 13:56

POC Alert All

dns

vsys1

TapZone2

TapZone2

55305

53

udp

alert

Suspicious DNS Query (P2P-Worm.palevo:brero.balkan-hosting.net)(4022349)

any

medium

client-to-server

12/3/2017 13:56

spyware

12/3/2017 13:56

POC Alert All

dns

vsys1

TapZone2

TapZone2

55305

53

udp

alert

P2P-Worm.palevo:brero.balkan-hosting.net(3839431)

any

medium

client-to-server

12/3/2017 13:56

spyware

12/3/2017 13:56

POC Alert All

dns

vsys1

TapZone2

TapZone2

55305

53

udp

alert

Suspicious DNS Query (P2P-Worm.palevo:brero.balkan-hosting.net)(4022349)

any

medium

client-to-server

I really don't know how to close this case. Who is right and how to prove either of the anti-malware wrong?

The workstation in question is a Windows 7 machine, the whole network is isolated (meaning no Internet access). The DNS server is Windows 2012, it is AD-integrated, no access to Internet. The workstation has McAfee VSE 8.8 and is definitions updated everyday. It also has McAfee HIPS, Solidcore, DLP and RSD.

[jvalentine]

If this is something that is happening repeatedly, you could use a tool like "Microsoft Message Analyzer" to try and map the process on the Windows 7 system making the DNS query for the suspicious domain.  

 - https://www.microsoft.com/en-us/download/details.aspx?id=44226  (download link)

 - https://technet.microsoft.com/en-us/library/jj649776.aspx  (operating guide)

I will say that the Microsoft Message Analyzer is a very powerful tool (translate: difficult to use).  In fumbling around with it I was able to map a DNS lookup (mail server) to the requesting process (outlook.exe) - but don't ask me how I did it.  

If anyone has a better method of mapping a client DNS lookup to a specific process, I'd love to hear about it.  Good luck.  

[SingChung]

Thanks. I had actually used Microsoft Message Analyzer to try to capture any DNS query traffic but after 3 days of sniffing I wasn't able to capture anything, no dns query at all. So, there must be something I didn't do correctly or the software need something special action.

[BPry]

@SingChung,

One thing that I would look at in the logs is what exactly the URL is that they are trying to access. I would say that most of the alerts that I recieve from this are generated by the user navigating to a website with less than perfect advertisement sources. If that doesn't work then Message Analyzer is probably your best bet, but it is a very very difficult tool to actually understand how to use. 

If all else fails make the user use another machine temporarly, say a day or two, so that you can 'look' at their current machine. See if the alerts change to the new machine or not; that would at least tell you if it's something they are doing or if it's truly a machine action. 

[SingChung]

BPry,

From the log, I guess these are the URLs:

brero.balkan-hosting.net

banjalucke-ljepotice.ru

The enironment is isolated, the user won't be able to go Internet. Therefore it is unlikely that the user has access website that contains hiding malicious advertisements.

The machine is already put offline and user not using it. I am currently working on a virtual clone of the machine in my Hyper-V environment. So, far I can't find anything suspicious on the machine.

Is it that there is no possibility the Palo Alto device is generating false positives?

[SingChung]

I analyzed the user logon and logoff events, I notice something interesting.

The Palo Alto detection has some relationship with user logon events. Whenever there is a interactive logon (at console or rdp), it was followed by a 3 minutes detection of malicious URL dns queries. The machine itself we can safely say no malware. The funny part is why there is always a detection for 3 minutes of same pattern whenever a user logs on. A check in Windows scheduler, these are programs that are scheduled to launch whenever a user logons on: Lenovo Solution Center, Lenovo Message Center Plus, Office Telemetry Agent, Realtek Lenovo MICPKey, Realtek Control Panel.

If user logs on but logoff in 2 minutes, it won't detect such malicious query (probably because the scheduled programs haven't finished starting up). So, it is evidence that one or more of these logon starting programs is doing some dns queries when starting up (perhaps to check for updates) but Palo Alto device read them as malicious. 

Is there a way for me to prove if the detections were indeed false positives?

[BPry]

@SingChung,

You would need to capture the URL requests that the machine is actually using something like Fiddler2 or a debugging proxie like Charles Web Debugging Proxy. I would say that it probably isn't a false positive. It's pretty simple for the Palo Alto to read the URL that was requested, so you don't usually actually see false positives with URL detection. 

[SingChung]

I managed to capture the malicious dns query. I enable the dns client event logging, logged off, then log on. The event id 1015:

_______________________________________________________________________________________________________________

- System

- Provider

[ Name] Microsoft-Windows-DNS-Client
[ Guid] {1C95126E-7EEA-49A9-A3FE-A378B03DDB4D}

EventID 1015

Version 0

Level 4

Task 0

Opcode 0

Keywords 0x8000000000000000

- TimeCreated

[ SystemTime] 2017-12-26T02:28:55.443725400Z

EventRecordID 712

Correlation

- Execution

[ ProcessID] 1468
[ ThreadID] 4328

Channel Microsoft-Windows-DNS-Client/Operational

Computer (removed due to confidentiality)

- Security

[ UserID] S-1-5-20

- EventData

QueryName kreten.banjalucke-ljepotice.ru
AddressLength 28
Address 170000350000000024063003200200055AEF68FFFE249F2600000000

________________________________________________________________________________________

I can track down to process DNS Client as the querying process. unable to track which process asked the DNS client to do the query.

From the timing of the query, and the running of the scheduled tasks triggered by user logon, there are only these possibilities:

1. HotStart

2. CacheTask

3. OfficeTelemetryAgentLogOn

4. SystemSoundsService

All these are Microsot system tasks.

[BPry]

@SingChung,

Scheduled tasks can be hijacked fairly easily. I would actually audit the task and make sure that they don't include a new call to this domain; my guess would be that the task itself has been hijacked and is now doing something with that domain. 

[BPry]

@SingChung,

You might want to check out this as well. It directly references that URL, along with showing the Reg keys that it creats in Winlogon. Might want to verify that it isn't what you are looking for. 

https://www.sophos.com/en-us/threat-center/threat-analyses/viruses-and-spyware/W32~Palev-Gen/detaile...