KB5126836 (Oct 2026 Monthly Rollup, Server 2012 R2) – all FC/MPIO SAN disks fail to start (Code 10) after reboot

Jayadeep Reddy Padala 25 Reputation points
2026-10-26T03:36:18.52+00:00

Has anyone else seen SAN disks disappear after installing KB5126836 on Windows Server 2012 R2?

Environment:

  • Windows Server 2012 R2 Standard (ESU), Dell PowerEdge R640
  • Emulex LPe35002 FC HBAs (elxfc 12.0.257.9, firmware 14.0.539.26)
  • Microsoft MPIO with MSDSM, Dell EMC array (DGC VRAID), ALUA
  • 3-node Oracle RAC cluster using the SAN LUNs for ASM

What happens:

After installing the October 2026 updates (KB5126836, KB5126773, KB5127356, KB5127688) and rebooting:

  • All SAN disk paths ("DGC VRAID SCSI Disk Device") show "This device cannot start (Code 10)" in Device Manager.
  • MPIO reports no disks (mpclaim -e: "There is no enterprise storage connected").
  • Disk Management only shows the local boot disk.
  • System log at boot: mpio Event 1 (MPIODisk created), then Event 46 (path removed due to a PnP event) and Event 17 (degraded) for every disk in the same second.

What is fine:

  • Both HBA ports are online, each sees the storage ports, and all LUN paths are discovered.
  • MPIO claims (DGC VRAID), MPIO settings and driver configuration are identical to working servers.

What changed:

KB5126836 installs a new disk class driver package (disk.inf 6.3.9600.23447), replacing disk.sys, classpnp.sys (6.3.9600.23447) and partmgr.sys. setupapi.dev.log shows it updating all SAN disk devices from 6.3.9600.18758 to 6.3.9600.23447.

Why we think it is the update:

  • On one node, the SAN disks were fine after the first patch reboot. They failed after the next reboot, when the Monthly Rollup finished and the disk drivers were updated.
  • On another server with identical hardware, HBA driver and firmware, the rollup failed to install due to Azure arc connectivity failure and rolled back. The SAN disks failed while it was applied and came back as soon as it rolled back.
  • A server that does not have the October updates works normally.

We couldn't find any known issue for this from Microsoft.

Questions:

  1. Has anyone seen KB5126836 (or the new disk.sys/classpnp.sys) break FC/MPIO SAN disks on 2012 R2?
  2. Has anyone seen it with QLogic HBAs as well, or only Emulex?
  3. Did anyone find a fix other than removing the update (for example, a newer HBA driver)?

Thanks in advance.

Windows for business | Windows Server | Storage high availability | Other
0 comments
Answer accepted by question author
Daphne Huynh (WICLOUD CORPORATION) 2,540 Reputation points Microsoft External Staff Moderator
2026-10-26T04:08:46.0166667+00:00

Welcome to Microsoft Q&A!

Thank you for taking the time to document the environment, event sequence, driver changes, and especially the rollback comparison in such detail

Based on the behavior you described, there is a clear correlation in your environment between installing KB5126836 and the FC/MPIO SAN disks becoming unavailable after restart.

In particular, the fact that the LUN paths remain visible at the HBA level while the disks subsequently fail with Code 10 and disappear from MPIO warrants deeper investigation of the Windows storage stack after the update.

However, the public documentation for KB5126836 does not currently document this specific FC/MPIO SAN-disk behavior as a known issue or provide a corresponding fix.

It is a recommendation verifying the storage paths, HBA status, zoning, drivers and firmware, MPIO configuration, Device Manager status, and relevant event logs when disks or MPIO paths disappear after a restart.

  • Possible cause

Given the timing you documented and the fact that SAN access returned after the update was rolled back, KB5126836 is an important correlation to investigate.

However, based on the currently available public documentation, it would not be appropriate to conclude that disk.sys, classpnp.sys, the Emulex HBA driver, or another specific component is definitively responsible. The available public information also does not establish whether this behavior is limited to Emulex HBAs or can occur with QLogic HBAs.

Confirming that distinction requires analysis of diagnostic data from an affected system rather than additional configuration comparisons through the forum.

  • Recommendation

Since rolling back KB5126836 has already restored SAN access in your testing, I recommend keeping the affected production systems at the last known-good state while the issue is investigated, subject to your organization's normal change-management, backup, recovery, and security requirements.

Because these LUNs are used by an Oracle RAC cluster, I would avoid making additional storage-stack or driver changes simply for troubleshooting unless they are part of a controlled test or recommended by Microsoft Support.

  • Next validation step

At this point, I strongly recommend opening a Microsoft Support ticket.

I appreciate the amount of troubleshooting you have already done. Normally, I would continue narrowing down the issue here, but this particular scenario requires a deeper level of investigation to determine whether the behavior is related to the update or to an interaction within the storage stack.

Microsoft Q&A is a community support forum, so there is a practical limit to how far this type of issue can be investigated here. The MPIO troubleshooting guidance recommends contacting Microsoft Support when the documented troubleshooting does not resolve the issue and identifies diagnostic information that can be collected for further investigation.

I understand that being referred to Microsoft Support may not be the answer you were hoping for, especially after the considerable time and effort you have already spent isolating the issue. However, continuing troubleshooting through the forum would likely repeat many of the configuration checks you have already completed without providing a reliable root-cause determination.

For this reason, opening a Microsoft Support case is the most appropriate next step. This will allow the issue to receive the deeper investigation it requires through the appropriate support channel.

References:

October 13, 2026—KB5126836 (Monthly Rollup) | Microsoft Support

Multipath I/O (MPIO) Troubleshooting Guidance - Windows Server | Microsoft Learn

If you find it useful, please click Accept Answer.

Thank you for choosing Microsoft Q&A.

Was this answer helpful?

3 people found this answer helpful.
0 comments

2 additional answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. Jayadeep Reddy Padala 25 Reputation points
    2026-10-26T10:09:41.4266667+00:00

    Thanks @Daphne Huynh (WICLOUD CORPORATION) , appreciate the detailed response.

    A quick update for anyone who finds this thread later:

    We uninstalled KB5126836 from one of the affected nodes and restarted it. All SAN disks came back straight away:

    • MPIO shows all 8 disks again, each with both paths (Active/Optimized and Active/Unoptimized), the same as our unaffected servers.
    • The disks are online and writable.
    • The disk driver is back to 6.3.9600.18758.
    • At boot, the MPIO events now show only the normal "disk created" and "device added" events, with no path removals or degraded warnings.

    We also compared the disk, MPIO and HBA driver configuration between an affected server and an unaffected server with identical hardware. The settings are the same, and the only difference is the version of the Windows disk drivers installed by KB5126836.

    So in our environment (Windows Server 2012 R2, Emulex LPe35002 FC HBAs, Microsoft MPIO/MSDSM, Dell EMC storage), KB5126836 is confirmed as the trigger.

    As ESU for 2012 R2 ended on 13 Oct 2026, these servers are not under Microsoft support anymore, so we're keeping the update off these servers. I'll update this thread if we find anything else.

    Was this answer helpful?

    3 people found this answer helpful.
    0 comments

  2. AI answer

    2026-10-26T03:37:25.7266667+00:00

    The only currently supported recovery for the affected RAC nodes is to remove KB5126836 during a controlled maintenance window, because Microsoft has not published a fix or mitigation for this specific Server 2012 R2 FC/MPIO Code 10 regression.

    What is confirmed

    • KB5126836 applies to Windows Server 2012 R2 ESU and is the final update of that ESU program.
    • Its published known-issues section lists a USB Audio Class 1.0 Code 10 issue, but no FC, MPIO, MSDSM, disk.sys, classpnp.sys, or SAN-disk issue. Microsoft therefore has not publicly classified this symptom as a known KB5126836 problem. ^1^
    • The evidence described—failure after the class-driver update, recovery immediately after update rollback on equivalent hardware, and an unpatched control server remaining operational—is strong A/B evidence of an update interaction. It does not establish which updated binary or interaction is responsible.
    • There is no published evidence here that QLogic adapters are affected, or that the issue is Emulex-specific.
    • No documented later hotfix, superseding Server 2012 R2 update, or HBA-driver version is identified as correcting this exact issue.

    Immediate containment and recovery

    1. Stop deployment of KB5126836 to the remaining RAC nodes.
    2. Move or stop RAC workload cleanly before changing an affected node. Do not manipulate ASM LUNs, disk signatures, partitions, or reservations while Windows cannot start the disk device objects.
    3. Capture evidence before rollback:
      • System and Setup event logs.
      • %windir%\inf\setupapi.dev.log.
      • msinfo32 system information.
      • mpclaim -e and mpclaim -s -d output.
      • Device Manager status and hardware IDs for several affected DGC VRAID disk objects.
      • File versions and hashes for disk.sys, classpnp.sys, partmgr.sys, mpio.sys, msdsm.sys, and the Emulex miniport.
    4. Uninstall KB5126836 from one affected node through Control Panel → Programs and Features → View installed updates, then reboot in the approved maintenance window.
    5. Confirm that:
      • The disk devices no longer report Code 10.
      • mpclaim -s -d again lists the LUNs and paths.
      • Disk Management sees the expected disks.
      • ASM sees the expected devices before returning the node to service.
    6. If rollback restores the LUNs, retain that node as the known-good baseline and do not reinstall the rollup until Microsoft and the HBA/storage vendor provide a validated resolution.

    Rollback removes security fixes, so isolate and protect the rolled-back systems while the incident is investigated.

    HBA-driver test

    Microsoft’s MPIO guidance recommends current, vendor-supported storage, HBA, multipath drivers, and firmware when path loss or Event 46 occurs. ^2^

    Treat a newer Emulex driver as a conditional lab test, not an established fix:

    1. Ask the HBA vendor for a driver/firmware pair explicitly supporting:
      • LPe35002
      • Windows Server 2012 R2
      • Microsoft MPIO/MSDSM
      • The installed Dell EMC array and ALUA configuration
    2. Test that pair on a non-production host first with KB5126836 installed.
    3. Exercise multiple cold boots and path failover/failback before considering RAC deployment.

    Do not independently mix arbitrary driver and firmware versions. Public QLogic/Emulex download availability does not prove that a package supports this legacy OS and exact storage stack. ^3^

    What not to pursue first

    Because the FC ports see the targets and every disk PDO fails with Code 10 immediately after the class-driver transition:

    • Rechecking zoning, MPIO claims, and ALUA settings would repeat checks already completed.
    • san policy=OnlineAll cannot resolve a disk device that fails to start.
    • Reinstalling MPIO, removing hidden disk devices, or running mpclaim -r -i -a across a production RAC node could change device enumeration without addressing the demonstrated update boundary.
    • Do not run chkdsk, initialize disks, clear reservations, or modify ASM LUN metadata.

    Escalation

    Open incidents with Microsoft Support for Business and the HBA/storage vendor. Provide the paired before/after evidence and request confirmation of:

    • Whether build 6.3.9600.23447 introduced a disk-class/Storport/MPIO compatibility regression.
    • Whether a private mitigation or corrected binary exists.
    • Whether Emulex LPe35002 driver [IP Address] is validated with that build.
    • Whether any QLogic miniports reproduce the same failure.

    Windows Server 2012 R2 ESU ended on October 13, 2026, so a supported-OS migration should proceed in parallel; KB5126836 is the final ESU security update. ^1^

    References

    1. October 13, 2026—KB5126836 (Monthly Rollup) | Microsoft Support
    2. Multipath I/O (MPIO) troubleshooting guidance
    3. Support | Public Driver Downloads | QLogic Fibre Channel HBAs and ...
    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

    2 people found this answer 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.