virtualbox.org

We have windows server 2008 R2 enterprise. 256 Gb memory. Each VM has 6 Gb memory (each guest is windows).
While running VirtualBox on the server we faced strange behavior. It's possible to run 22 VMs, but when one starts 23rd VM (let's name it 'VPC23'), it starts and then stops with error (below).
But, the interesting thing is, when you power off some of the running VMs and start that previously 'non starting' VM (VPC23), it starts smoothly.
In short, we can run 22 machines simultaneously, but not 23 or 24.
Are there limitations on the number of concurrently running VMs? Or are there some limitations with memory? Or is it windows issue?

The virtual machine 'xxx' has terminated unexpectedly during startup with exit code 1 (0x1). More details may be available in 'C:\VMs\XXX\Logs\VBoxHardening.log'.
E_FAIL (0x80004005)
MachineWrap
IMachine {85632c68-b5bb-4316-a900-5eb28d3413df}

Hardenning log:
1c80.1c84: KiUserExceptionDispatcher: 0xc0000005 (0000000000000001, 0000000000000024) @ 0000000076d92b04 (flags=0x0)
rax=0000000000000000 rbx=0000000076e92ce0 rcx=00000000fffffffc rdx=000000000000000c
rsi=000000000000000c rdi=0000000000000000 r8 =000000000022c1e8 r9 =0000000000000004
r10=0000000000000000 r11=0000000000000246 r12=0000000076e92d30 r13=0000000000000000
r14=000007fffffde000 r15=0000000000000000 P1=000000000022dc60 P2=0000000000000001
rip=0000000076d92b04 rsp=000000000022c230 rbp=0000000000000000 ctxflags=0010005f
cs=0033 ss=002b ds=002b es=002b fs=0053 gs=002b eflags=00010213 mxcrx=00001f80
P3=00000000ffffffdd P4=000000013fcfa5ad P5=000000013fcfa3f0 P6=000000000022bb50
dr0=0000000000000000 dr1=0000000000000000 dr2=0000000000000000 dr3=0000000000000000
dr6=0000000000000000 dr7=0000000000000000 vcr=0000000000000000 dcr=0000000000000000
lbt=0000000000000000 lbf=0000000000000000 lxt=0000000000000000 lxf=0000000000000000
1558.1c70: Error -35 in suplibOsInit! (enmWhat=1)
1558.1c70: Integrity error (0xe986ffdd/-35): NtQuerySystemInformation/SystemExtendedHandleInformation failed: 0xc000009a
1c34.1c38: supR3HardNtChildWaitFor[1]: Quitting: ExitCode=0x1 (rcNtWait=0x0, rcNt1=0x0, rcNt2=0x103, rcNt3=0x103, 478 ms, the end);

Please supply the full log, not a snippet,
What version of VB are you using?
What patch level is W2008R2?

Be aware that this forum is for the free VB version, so please let us know if you are using the free one or if you have a VB Enterprise license.

This looks like a bug in the VirtualBox code, which calls NtQuerySystemInformation(SystemExtendedHandleInformation) and handles the STATUS_INFO_LENGTH_MISMATCH (0xc0000004) return code, but not STATUS_INSUFFICIENT_RESOURCES (0xc000009a). Please create a ticket in the Bugtracker.

This isn't something obvious. Wouldn't think STATUS_INSUFFICIENT_RESOURCES is returned when an increased buffer would help. Doesn't match the general logic (and all available code snippets I could find which are using NtQuerySystemInformation).

Could you provide VBox.log from the last successfully started VM and the full hardening log file (of the first failed start)? Would give us a lot of additional detail.

In a bug report or here...

VirtualBox version 6.1.30 r148432 (Qt 5.6.2)
Guest Machines are all windows 7 x64 with 6 Gb of memory.
Here are logs for the specific VM which can be started while it's 22nd, and can't when it's 23rd.
(In fact it doesn't matter which machine can or can't start. It matters only whether it 23rd or not)

Successful start
https://www.mediafire.com/file/t38s7g1r ... s.zip/file
Failed start
https://www.mediafire.com/file/2v2yrjyn ... r.zip/file

Case can be closed. It's windows 2008 memory management issue.
The issue was resolved after installing Windows 10 corp on the same hardware. And it became possible to run more than 22 machines.