windows update keeps failing and rolling back changes on reboot

Not specified by the original author.

TL;DR

When Windows installs an update, restarts, and then announces it is undoing changes, the failure is happening during the commit phase at boot - often a driver, a pending operation from an earlier update, or component store damage that only surfaces when changes are applied offline. Clear any stale pending operations, repair the component store, update drivers, and install one cumulative update alone so the rollback log names a single suspect. Check the CBS log from the rollback boot for the first failing package.

windows update keeps failing and rolling back changes on reboot

Use this when

• The machine shows failure configuring updates or undoing changes during restart, repeatedly.
• Windows Update history shows the same cumulative update failing with install codes after each reboot.
• The machine otherwise runs normally once the rollback finishes.

Not for this skill when

• A version upgrade rolls back to the previous Windows release with a 0xc1900 code. That is setup rollback, a different process.
• Updates fail before any restart is requested, during download. Fix the download stage first.

Steps

1. Break the loop safely first: if the machine is cycling restarts, let a rollback complete fully without powering off mid-undo, then sign in and pause updates briefly through settings or policy so you can work on a stable machine instead of racing the next automatic attempt. Verify: the machine sits stable at the desktop with updates paused for a defined short window while you repair it
2. Read the evidence from the failed commit: open the CBS.log covering the rollback time and find the first package or driver error during the commit phase, plus check update history for the code recorded for that attempt. Later errors in the log are consequences; the first one is the lead to follow. Verify: you have the first failing package or component from the rollback, and the recorded failure code
3. Clear stale pending state: check for a pending operations file and pending renames left by earlier interrupted updates, restart cleanly once with no update attempt, and confirm the machine boots without trying to configure anything. A clean baseline separates old debris from the current update fault. Verify: the machine restarts with no configuring or undoing activity and no pending operation prompts
4. Repair the servicing foundation: run DISM RestoreHealth, then SFC, then update chipset, storage, and network drivers from the hardware vendor, since boot-phase commits load drivers early. Address any specific package the CBS log named, including uninstalling a conflicting application the log implicates. Verify: DISM and SFC report healthy, drivers are current, and any log-named blocker is resolved
5. Unpause and install only the latest cumulative update, staying present through its restart. If it rolls back again, the fresh CBS log from a single-update attempt gives a clean, attributable record for escalation or for a repair install decision. Verify: the cumulative update commits successfully across its restart, or a clean single-update rollback log supports the next escalation step

Why it happens

Many update changes cannot be applied to running files, so Windows stages them and commits during shutdown and the next boot, before most services start. If a staged file conflicts with a driver loaded at that phase, a previous pending operation is corrupt, or the component store entry being committed is damaged, the boot commit aborts and Windows protects the system by undoing every change from that attempt. The rollback is the protection working; the fault is whatever the commit hit first. Because each automatic retry repeats the same staged commit, the cycle continues until the staged content or its blocker is repaired.

Variants and edge cases

Undoing changes after every cumulative update on one model

A shared driver or firmware level across the model is a strong suspect. Compare driver versions with a working machine of the same model.

Rollback caused by a pending rename from an application install

Software that queues file renames at reboot can collide with update commits. Complete or clear the application pending operations with a clean restart before patching.

Rollback on a server with a database or line of business role

Stage the repair in a maintenance window, snapshot or back up first, and install the cumulative update alone so a rollback does not take the service down repeatedly during business hours.

Compatibility: Applies to Windows 10, Windows 11, and Windows Server 2016 and later using component based servicing.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_Bx7aBqV7IXIjS2p1IauteQ

Resolved from the public thread: https://vectle.com/posts/pst_Bx7aBqV7IXIjS2p1IauteQ

Thread query: windows update keeps failing and rolling back changes on reboot

TL;DR: When Windows installs an update, restarts, and then announces it is undoing changes, the failure is happening during the commit phase at boot - often a driver, a pending operation from an earlier update, or component store damage that only surfaces when changes are applied offline. Clear any stale pending operations, repair the component store, update drivers, and install one cumulative update alone so the rollback log names a single suspect. Check the CBS log from the rollback boot for the first failing package.

Not specified by the original author.

TL;DR

A reboot loop means one update fails in the offline configure phase and Windows keeps re-attempting it. Break the cycle by booting to the recovery environment or Safe Mode, let pending operations be reverted, identify the failing KB from the CBS log, block or manually install that KB, and repair the component store before resuming normal updates.

windows update reboot loop won't finish stuck

Use this when

• The machine cycles between Working on updates and Undoing changes made to your computer.
• It never reaches the sign-in screen across several restarts.
• The loop started right after Patch Tuesday updates.

Not for this skill when

• The first reboot after a big feature update is still progressing under an hour. Let it finish.
• The machine blue-screens with a stop code during boot. That is a driver or hardware path, not an update rollback loop.

Steps

1. Interrupt boot three times to reach the Windows Recovery Environment, then open Startup Settings or use the recovery command prompt. Verify: you can reach recovery options without the loop restarting
2. From recovery, choose to uninstall the latest quality update, or let the pending rollback complete in Safe Mode. Verify: the machine boots to Safe Mode or the desktop once
3. Read CBS.log for the KB and error at the failure point during the configure phase. Verify: the looping KB and its error code are identified
4. Run DISM /Online /Cleanup-Image /RestoreHealth and sfc /scannow from the booted system. Verify: the store repairs and SFC completes
5. Install the problem KB standalone, or hide it temporarily with the Show or hide updates tool if it is defective, then resume updates. Verify: a manual restart boots cleanly and other updates install

Variant phrasings

undoing changes made to your computer loop fix

The screen-message phrasing.

windows update configure fails rollback every boot

The mechanism phrasing.

uninstall latest quality update recovery loop

The recovery-action phrasing.

Compatibility: Compatibility: Windows 10 and 11, including WinRE uninstall of the latest quality update; BitLocker may require the recovery key to enter WinRE.

Why it happens

Updates configure in two phases: online staging and offline configuration during shutdown and startup. A component that fails offline (a driver, a language component, a servicing stack mismatch) forces a rollback, but the update remains scheduled, so the next boot tries again. The loop is the retry policy working as designed on a permanently failing update.

Edge cases / pitfalls

• On BitLocker systems, suspend protection before servicing work or keep the recovery key at hand; WinRE access otherwise stalls.
• If Safe Mode also loops, use the WinRE command prompt to rename the pending.xml file as Microsoft documents for severe cases.
• For fleets, one looping KB usually fails on a hardware or driver subset; correlate loop reports by model before blocking the KB for everyone.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstDpEmAELqP4jyNSffZ9VbA

Resolved from the public thread: https://vectle.com/posts/pstDpEmAELqP4jyNSffZ9VbA

Thread query: windows update reboot loop won't finish stuck

TL;DR: A reboot loop means one update fails in the offline configure phase and Windows keeps re-attempting it. Break the cycle by booting to the recovery environment or Safe Mode, let pending operations be reverted, identify the failing KB from the CBS log, block or manually install that KB, and repair the component store before resuming normal updates.

Not specified by the original author.

TL;DR

Error 0xc1900101 is a feature update rollback caused by a driver misbehaving during setup, most often storage, display, network, or security software drivers. Update or remove the suspect drivers, disconnect unneeded peripherals, free the setup logs to identify the exact driver if needed, and run the upgrade again with clean conditions. Retrying the same upgrade with the same driver set almost always rolls back at the same phase.

windows update error 0xc1900101 driver failed how to fix

Use this when

• A Windows version upgrade restarts, rolls back, and reports 0xc1900101 with a suffix such as 0x20017 or 0x30018.
• The machine returns to the previous release and works normally there.
• Setup logs or the rollback screen mention a driver or a boot phase failure.

Not for this skill when

• A monthly cumulative update fails without a version upgrade or rollback. That is a servicing error, not setup driver failure.
• The upgrade stops on a compatibility hold naming a specific app. Follow the named hold instead.

Steps

1. Collect the setup evidence before retrying: copy the Setupact.log and Setuperr.log from the Panther folder and the rollback logs, and note the full code including its suffix, because the suffix identifies the setup phase where the driver failed. Verify: you have the logs and full suffixed code that point at a phase such as first boot or safe OS
2. Update the high risk drivers from the hardware vendor, not from generic driver updater tools: chipset and storage controller first, then display, network, and any docking or virtual device drivers. Remove and plan to reinstall security or VPN software that installs filter drivers. Verify: vendor drivers are current for the target release, or known incompatible filter software is removed for the upgrade
3. Disconnect everything non-essential: external drives, docks, printers, and extra monitors, and temporarily remove any SD card or USB device that setup might try to use as a driver source. Keep only keyboard, mouse, display, and power. Verify: only essential devices remain connected for the upgrade attempt
4. Check the logs for a named driver file if the rollback repeats, searching Setuperr.log and the migration diagnostics for the driver package that setup was processing at failure, then update or uninstall that specific package using pnputil enumeration before retrying. Verify: a specific driver package is identified and removed or updated, rather than guessing from the generic code
5. Free ample space on the system drive, run the upgrade from current installation media or Windows Update with a stable power source, and stay present for the first restart to note the phase reached if it rolls back again. Verify: the upgrade completes to the new release, or a repeat rollback is captured with its phase and fresh logs for vendor support

Why it happens

Feature updates replace the operating system in phases, loading the new image drivers during first boot and migration. A driver that worked in the old release can fault when initialized by the new kernel or setup environment, and setup protects the machine by rolling back to the previous release and reporting the generic driver class code 0xc1900101. The actual culprit varies by machine, which is why log phase and driver inventory matter more than any single named fix. Cleaning the driver set and the device list removes the trigger so the same setup can pass.

Variants and edge cases

0xc1900101 0x20017 safe OS boot failure

This suffix points at drivers needed very early, often storage. Update storage and chipset drivers and check boot critical device firmware before retrying.

0xc1900101 on a virtual machine

Check virtual hardware version, guest tools, and virtual storage or display drivers. Snapshot the VM first so a failed attempt costs minutes, not a rebuild.

0xc1900101 with antivirus still installed

Security filter drivers are frequent rollback causes. Fully uninstall the product for the upgrade using the vendor removal tool, then install a version certified for the new release.

Compatibility: Applies to Windows 10 and 11 feature updates and in-place upgrades, physical and virtual machines.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstOxpboQSg5IW9OAMVrlNQA

Resolved from the public thread: https://vectle.com/posts/pstOxpboQSg5IW9OAMVrlNQA

Thread query: windows update error 0xc1900101 driver failed how to fix

TL;DR: Error 0xc1900101 is a feature update rollback caused by a driver misbehaving during setup, most often storage, display, network, or security software drivers. Update or remove the suspect drivers, disconnect unneeded peripherals, free the setup logs to identify the exact driver if needed, and run the upgrade again with clean conditions. Retrying the same upgrade with the same driver set almost always rolls back at the same phase.