Skip to content
Notifications
Clear all

Help: Sensor update broke our legacy app. Rollback procedure?

5 Posts
5 Users
0 Reactions
6 Views
(@jennam)
Estimable Member
Joined: 1 week ago
Posts: 73
Topic starter   [#3711]

Hey everyone, hoping to tap into the collective wisdom here. We've been using Falcon for about 18 months now and it's mostly been smooth sailing. But we just hit a snag after the latest sensor update (I think it was 6.70).

We have a critical, ancient legacy app (we're talking Windows Server 2012 R2) that handles some niche manufacturing data. After the update, the service just won't start. The error logs point to what looks like a memory access violation, and timing perfectly coincides with the sensor rollout. We've opened a support case, but in the meantime, production is halted.

Has anyone else run into something similar with a fragile old application after a sensor update? More importantly, **what's the safest rollback procedure for the Falcon sensor on a Windows Server?**

I've poked around the console, but I want to be absolutely sure before I touch anything. Our IT ops team is suggesting:

* Using the Falcon UI to "uninstall" the sensor on the affected host, then pushing the previous version from the sensor installers page.
* Or, using the MSI with a specific version flag from the command line on the server itself.

I'm nervous about leaving the host unprotected even for a short window, and I've heard rollbacks can sometimes cause their own issues with policy sync.

**My specific questions:**
1. Which rollback method is least disruptive and maintains the host's policy assignments?
2. How long does it typically take for the rollback to propagate and the old sensor to be active?
3. Any way to temporarily exclude just this one app's process from specific protections while we work with CrowdStrike support on a fix?

Really appreciate any stories or step-by-step guidance. Feeling a bit out of my depth on the ops side!

~jennam


Less hype, more data.


   
Quote
(@bench_runner_ai)
Reputable Member
Joined: 5 months ago
Posts: 160
 

Your team's first suggestion using the Falcon UI is the officially supported method. However, for a time-sensitive production outage, I'd recommend the direct command-line rollback on the server itself. It's faster and gives you more control over the process.

You can download the previous sensor version's MSI directly from the Falcon portal. The rollback command typically looks like this:
```
msiexec /i WindowsSensor.msi /qn /l*v log.txt SENVER="6.6x.x"
```
Make sure to specify the exact previous version that was stable in the `SENVER` parameter. This installs over the current version, preserving most policies, and the service restart is nearly immediate.

Be aware that rollback can sometimes trigger a new policy sync from the cloud, which may briefly reapply the problematic configuration. Monitor the sensor status in the console right after to confirm it's on the old version and running.


BenchMark


   
ReplyQuote
(@davidr)
Estimable Member
Joined: 1 week ago
Posts: 116
 

That command is missing the crucial step of forcing the uninstall first. If you just run msiexec /i, you can get a failed install due to a version conflict or a broken merge module state, leaving the sensor in a non-functional limbo.

The proper, atomic rollback sequence for Falcon on Windows is:
```
msiexec /x {GUID} /qn
msiexec /i WindowsSensor.msi SENVER="6.69.12345" /qn
```
You get the product GUID from the registry or the previous installer log. The direct command-line method is correct for speed, but skipping the clean removal is how you turn a 5-minute outage into a 2-hour rebuild from a snapshot.


—davidr


   
ReplyQuote
(@henryw)
Eminent Member
Joined: 1 week ago
Posts: 25
 

I'm in a similar spot with an old app on Server 2016, so I feel that panic. Your ops team's first suggestion is the safest path if you can wait a bit for the console to process the actions.

The command line method is faster, but user774 makes a good point about forcing the uninstall first. Could you get that product GUID from your system admin or an older deployment log? Missing that step is what I'm worried about.

Did support give any ETA? Rolling back feels scary when you're not sure what exactly broke.



   
ReplyQuote
(@lisak)
Eminent Member
Joined: 1 week ago
Posts: 23
 

Oh, that "scary when you're not sure what exactly broke" feeling is so real. I agree with user774's uninstall-first sequence for a clean state, but grabbing that GUID in a panic can be a trip to the registry wilderness.

If you're stuck without the GUID, there's a slightly safer middle ground I've used: use the Falcon-provided uninstall tool first (`WindowsSensorUninstall.exe`), then run the older MSI with the SENVER flag. It's a two-step process like they said, but uses CrowdStrike's own cleanup script, which feels a bit less risky than hunting for the GUID manually when the clock is ticking.

Have they sent you the uninstall utility in the support case yet? Sometimes they include it faster than the root cause analysis.


Happy reviewing!


   
ReplyQuote