On July 14, 2026, Microsoft rolled out KB5099536 for Windows Server 2025, bumping the OS build to 26100.33158. Within 24 hours, a troubling field report emerged: two servers equipped with LSI MegaRAID 9361 controllers and CacheCade enabled failed to return to normal operation after the update. The community account describes CacheCade volumes that disappeared and virtual drives stuck in an “Optimal access blocked” state. Microsoft has not yet listed this as a known issue, but the symptoms are severe enough that Windows Server administrators using similar hardware should act now — before the August Patch Tuesday deadline.

A Single Report, but a Serious One

The warning comes from a July 15 post on WindowsForum.com, where an administrator detailed the experience of two Windows Server 2025 machines with LSI 9361-series controllers and CacheCade acceleration enabled. After applying the July cumulative update, both servers hit problems: CacheCade volumes were missing, the controller showed blocked virtual drives, and the systems could not boot normally.

No other confirmed incidents have surfaced, and Microsoft’s official release-health dashboard for Windows Server 2025 makes no mention of MegaRAID, CacheCade, or boot recovery loops associated with KB5099536. The only acknowledged known issue for this update involves WSUS sync timeouts, a completely separate matter. The absence of a formal advisory means the problem is not yet confirmed, but it also means that administrators must rely on their own assessment rather than waiting for a Microsoft directive.

Who Is at Risk? Narrowing Down the Hardware Profile

The threat is not universal. Virtual machines, servers using other storage controllers, and Windows Server 2025 hosts without CacheCade are almost certainly unaffected by this specific interaction. The caution applies to physical servers that combine three elements:

  • Windows Server 2025 (the only platform in the report)
  • An LSI MegaRAID 9361-class RAID controller
  • CacheCade enabled (a feature that uses SSDs as a read/write cache for a virtual drive)

If your server matches this profile—particularly if the boot drive resides on the controller or the system lacks redundancy—you should treat KB5099536 as a high-risk deployment. Servers that merely use MegaRAID without CacheCade, or run newer controller models, probably fall outside the reported failure domain, but the evidence is thin enough that some extra validation is still wise.

Understanding KB5099536: More Than a Routine Patch

KB5099536 is the July 2026 cumulative update for Windows Server 2025, advancing the OS to build 26100.33158. Among its documented fixes is the resolution of a Recycle Bin filename bug that had persisted since June. Like any monthly rollup, it also bundles security patches and quality improvements. The update itself is not inherently dangerous, but the interplay between low-level storage drivers, controller firmware, and Windows boot code can expose edge-case regressions.

That the official known-issue list lags behind real-world reports is normal. Microsoft’s process requires gathering enough data to reproduce and diagnose a problem before it can be added to the public dashboard. Until then, the onus falls on IT teams to detect and contain potential issues early. The July 15 report is a credible early warning, not a confirmed disaster.

Preparing Your August Deployment: Three Paths Forward

August Patch Tuesday falls on August 11, 2026. Between now and then, administrators should sort their Windows Server 2025 fleet into three distinct groups:

  1. Standard patch cycle for systems that do not match the reported hardware. Continue with pilot rings and regular post-update checks. This includes virtual machines, hosts using different controllers, and servers without CacheCade.
  2. Temporary delay for physical machines with LSI MegaRAID 9361 and CacheCade enabled. Place these servers in a separate deployment ring. Document the hold with clear release criteria—such as a successful test on a representative box, an official statement from Microsoft or Broadcom, or enough uneventful production deployments to reduce uncertainty.
  3. Supervised patching for at-risk servers that must receive the July update urgently. Treat the installation as a storage maintenance event, not a routine reboot.

This tiered approach avoids the twin mistakes of freezing all updates across the board and ignoring a credible field report.

The Safe Patching Checklist for At-Risk Servers

If you cannot postpone KB5099536 for a CacheCade-enabled host, follow a rigorous validation plan:

  1. Inventory the storage hardware. Confirm the exact RAID controller model and firmware version. Check whether CacheCade is active by reviewing the controller’s management interface.
  2. Record the pre-update state. Document the controller configuration, virtual drives, CacheCade volumes, and all Windows-visible storage.
  3. Verify backups. Ensure you have a full, tested backup that can be restored independently of the controller being updated. A backup that relies on the same storage path offers no protection.
  4. Establish out-of-band access. Make sure you can reach the server through a remote management console (iDRAC, iLO, etc.) so you can intervene if Windows fails to start.
  5. Install the update during a maintenance window. Do not rely solely on centralized update status. Watch the entire reboot process from the console.
  6. Inspect the controller early in the boot. As the system restarts, enter the RAID configuration utility and confirm that virtual drives and CacheCade volumes are present and healthy.
  7. Confirm visibility in Windows. After the OS loads, verify that all expected disks and volumes appear in Disk Management and that applications can access their data.
  8. Perform a second reboot. A restart can reveal intermittent problems. If the first boot succeeds, boot again before declaring the update safe.
  9. Keep the server out of production if CacheCade volumes are missing, virtual drives show blocked access, or Windows enters automatic repair.

These steps cannot guarantee safety under every workload, but they capture the failure pattern described in the community report before user data is at risk.

When Things Go Wrong: Diagnosing a Failed Patch

If a server with the matching hardware profile fails to boot after KB5099536, avoid the reflexive urge to force Windows recovery or roll back the update. The root cause may lie in the storage layer, not the operating system.

  • Check the controller status first. Access the RAID firmware during POST. If virtual drives are marked as “Optimal access blocked” or CacheCade volumes are absent, the controller itself is in a problematic state.
  • Do not make speculative changes. Resetting the controller configuration or removing CacheCade assignments can destroy data. Gather exact error messages and the timeline of events before taking any action.
  • Document everything. Record the installation time, the first symptom, what appears in the controller logs, and whether rolling back the update (via Windows Recovery Environment or safe mode) restores normal operation. This detail is what can turn an anecdote into a reproducible bug for Microsoft or Broadcom.

Recovery loops—Windows repeatedly entering automatic repair—may be a downstream effect of a missing boot drive, not a problem with the update files themselves. Focus on why the storage is unavailable before trying to fix Windows.

The Road Ahead: Monitoring and Next Steps

Between now and August 11, administrators should:

  • Watch Microsoft’s release-health page and the KB5099536 support article for any new entries related to storage controllers.
  • Check Broadcom’s support portal for firmware advisories or compatibility updates for the LSI 9361 series.
  • Monitor the WindowsForum thread and other community channels for additional reports or successful deployments on similar hardware.
  • Test KB5099536 on a representative, non-production CacheCade system if one is available. A test that clones the exact storage path—controller model, firmware, CacheCade configuration—provides the strongest signal.

If Microsoft eventually confirms a compatibility issue, it may issue a Known Issue Rollback or an out-of-band fix. Until then, the evidence supports a narrow precaution: proceed normally for most Windows Server 2025 infrastructure, but ring-fence MegaRAID 9361 hosts with CacheCade, and patch them only after you can personally verify that the storage subsystem survives the reboot.