VPXD, consolidation, or backup failures in vCenter Server 8.0 U2 due to file lock information requests

Products

• VMware vCenter Server

Issue/Introduction

Symptoms:

• Recurring crashes of the VPXD service and VM consolidation or backup failures.
  • Analysis of service crash backtraces show entries similar to the following:

    #8  0x00005650a86c24f1 in VmMo::ConsolidateDisksStateLockedInt (this=this@entry=0x7f1944d6da90) at bora/vpx/vpxd/vm/snapshot.cpp:1484
    #9  0x00005650a86c2e2c in VmMo::ConsolidateDisks (this=0x7f1944d6da90, result=...) at bora/vpx/vpxd/vm/snapshot.cpp:1581

The above entries indicate vpxd is crashing when trying to fetch the file lock owners function:

• Disk consolidation or backup failures due to files being locked.
  • The following backtrace found in these cases shows this:

    (VPXD Backtrace):
    backtrace[08]: vpxd[Vmomi::MethodFault::Throw]   (Physical frame)
    [...]/bora/vim/lib/public/vmomi/MethodFault.h:170
    backtrace[09]: vpxd[VpxLRO::TrackAndWait]   (Physical frame)
    [...]/self/cwd/bora/vpx/common/vpxLro.cpp:1824
    backtrace[10]: vpxd[VpxLRO::TrackAndWait]   (Physical frame)
    [...]/self/cwd/bora/vpx/common/vpxLro.cpp:1724
    backtrace[11]: vpxd[VmMo::ConsolidateDisksStateLockedInt]   (Physical frame)
    [...]/cwd/bora/vpx/vpxd/vm/snapshot.cpp:1474
    backtrace[12]: vpxd[VmMo::ConsolidateDisks]   (Physical frame)
    [...]/cwd/bora/vpx/vpxd/vm/snapshot.cpp:1583

Environment

VMware vCenter Server 8.0.2

Cause

The crash is linked to a new feature implemented in vCenter Server 8.0 Update 2 to show file lock owner information in the UI, and failure messages.

This issue only happens when there is more than one virtual disk with file lock contention on the host.

Resolution

This issue has been fixed in vCenter Server 8.0 Update 2b.

Should you experience an issue that appears similar, despite have vCenter Server already updated to Update 2b, please reach out to the VMware Support by opening a support request, following the article Creating and managing Broadcom support cases

Workaround:
As of now, there are no work-arounds for this problem.