Skip to content
Notifications
Clear all

Help: Can't get the root cause analysis to work on a cryptolocker event

2 Posts
2 Users
0 Reactions
1 Views
(@crm_pragmatist)
Estimable Member
Joined: 2 months ago
Posts: 98
Topic starter   [#19262]

Alright, I'm hitting a wall and need some real-world troubleshooting, not marketing fluff. We had a cryptolocker event get through. Cybereason flagged the malicious process execution, which is fine, but the Root Cause Analysis (RCA) MalOp is garbage. It's just pointing to the initial binary that was blocked and calling it a day.

That's useless. I need the initial access vector. Was it a phishing link? A drive-by? An exploited vulnerability? The console's RCA view shows a shallow tree with maybe one or two parent processes, then dead ends. The "Related Malware" and "Suspicious Entities" panels are practically empty.

My setup:
* Sensor version 22.1
* Full environment telemetry enabled
* No custom exclusions that should affect this

What I've tried:
* Letting the MalOp "mature" for 24 hours – no change.
* Manually digging in the Process Tree and File Analytics – it's like the data isn't being correlated.
* Verified the endpoint wasn't in a stale/offline state during the incident.

Is the RCA feature just fundamentally broken for these kinds of fast, destructive attacks? Or is there a specific logging or config toggle everyone else is using that actually makes it work? I've been burned before by vendors where the demo "AI" works perfectly but the real-world product can't connect obvious dots.

What are you all seeing in your environments?



   
Quote
(@gracec)
Estimable Member
Joined: 1 week ago
Posts: 73
 

Yeah, I've seen this exact thing. The RCA is heavily dependent on having a rich, continuous data stream from before the trigger event. With a fast ransomware detonation, the initial vector might have been a scheduled task or a script that spawned, did its damage, and terminated all within a data collection cycle.

Your sensor version is a bit back. There were significant RCA correlation improvements in the 22.2 and 22.3 cycles specifically around linking spawned processes to parent scripts and network connections. Before you go deeper, check if your management console version is ahead of your sensor. A mismatch can break the analysis engine.

You mentioned manually checking File Analytics - did you look for any .LNK files, script files (PS1, VBS, JS), or emailed archives around that timestamp on that host? The console might not auto-correlate them, but the raw data is often there. It's a slog, but that's where I usually find the missing link.


The right tool saves a thousand meetings.


   
ReplyQuote