Windows 11 25H2 Devices Experiencing Servicing State Inconsistencies with Cumulative Updates

Smith, Colby A 20 Reputation points
2026-07-15T18:28:16.62+00:00

I'm looking for guidance on a Windows servicing issue that has been affecting a subset of devices in our environment for several months.

We manage approximately 3,000+ Windows 11 25H2 devices, and this issue is only occurring on a portion of them, roughly 100-300 devices. The remaining devices update normally and do not exhibit this behavior.

When reviewing Component Based Servicing (CBS) package states on affected systems, I consistently find that build 26100.6899 is marked as Superseded, build 26100.7840 is marked as Installed, and newer builds such as 26100.8037 and 26100.8246 are marked as Resolved. CBS logs also indicate that the RollupFix package for 26100.7840 is installed.

However, despite CBS reporting that 7840 is installed, the active operating system appears to still be running files from the older update level. For example, ntoskrnl.exe remains at version 10.0.26100.6899 on affected devices. This creates what appears to be a mismatch between the servicing state reported by CBS and the actual OS binaries that are active on the system.

Another consistent observation is that the newer update levels, such as 8037 and 8246, never seem to progress beyond the Resolved state. They are detected and considered applicable but never transition to Staged or Installed.

While reviewing CBS logs, I found entries indicating that TrustedInstaller had previously been interrupted:

"It looks like TI crashed or was interrupted last time it was started"

and

"Mark session as corrupt since it is neither complete nor pended"

CBS later appears to recover successfully and reports normal completion without an active reboot-pending state.

To troubleshoot, we have already performed the following actions on affected devices:

  • DISM CheckHealth
  • DISM ScanHealth
  • DISM RestoreHealth
  • SFC /SCANNOW
  • Reset Windows Update components by renaming both the SoftwareDistribution and Catroot2 folders
  • Multiple reboots
  • Repeated update deployment attempts through NinjaOne OS Apply

Despite these remediation steps, affected devices continue to show the same behavior:

  • 6899 marked as Superseded
  • 7840 marked as Installed
  • 8037 and 8246 marked as Resolved
  • ntoskrnl.exe remaining at 26100.6899
  • Newer cumulative updates never advancing beyond Resolved

Because this issue is impacting only a subset of devices across a large fleet and has persisted for months despite standard servicing remediation, I'm trying to determine whether this points to:

  • A CBS servicing state inconsistency
  • A partially committed cumulative update
  • A component store issue that DISM is not detecting
  • A known Windows servicing defect
  • Some other condition that can leave CBS reporting a newer RollupFix package as installed while the active OS binaries remain at an older revision

Has anyone encountered a similar situation where CBS reports RollupFix 26100.7840 as installed, while ntoskrnl.exe remains on 26100.6899 and subsequent cumulative updates remain stuck in the Resolved state? If so, what logs, package states, or remediation steps would you recommend investigating before considering an in-place repair upgrade?

Has anyone encountered a similar situation where CBS reports RollupFix 26100.7840 as installed, while ntoskrnl.exe remains on 26100.6899 and subsequent cumulative updates remain stuck in the Resolved state? If so, what logs, package states, or remediation steps would you recommend investigating before considering an in-place repair upgrade?

As a side note: Doing a clean install of windows has managed to fix these issues before however we would like to figure out if there is a more efficient solution.

Windows for business | Windows Client for IT Pros | Devices and deployment | Install Windows updates, features, or roles
0 comments No comments

Answer accepted by question author

Jason Nguyen Tran 24,530 Reputation points Independent Advisor
2026-07-16T00:30:08.1933333+00:00

Hi Smith, Colby A,

Based on your observations, this does look like a servicing state inconsistency where CBS is reporting a newer RollupFix as installed, but the active OS binaries remain at an older revision. This typically happens when TrustedInstaller or the servicing stack is interrupted mid-commit, leaving the component store in a partially updated state.

Since you’ve already run DISM and SFC without success, the next step I’d recommend is reviewing the CBS and Windows Update logs specifically around the time of the failed sessions. Look for entries tied to the RollupFix package that mention “corrupt session” or “rollback.” You may also want to check the Pending.xml file in the WinSxS folder, as leftover entries there can block newer updates from staging.

Another useful diagnostic is enabling advanced Windows Update logging (via registry) and tracing the servicing stack with Windows Performance Recorder. This can help confirm whether the issue is driver-related, a servicing stack defect, or a blocked commit. If you find that multiple cumulative updates remain stuck in the Resolved state, that’s a strong indicator the component store is not properly advancing.

At this point, if logs confirm repeated corruption or rollback, an in-place repair upgrade is the most reliable remediation short of a clean install. While disruptive, it preserves apps and data and resets the servicing stack. For large fleets, you may want to pilot this on a small set of affected devices first.

I hope the response provided some helpful insight. If you find this answer useful, please hit “accept answer” so I know it addressed your concern.

Jason.

Was this answer helpful?

1 person found this answer helpful.
0 comments No comments

0 additional answers

Sort by: Most helpful

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.