The issue was RESOLVED by completely disabling power-saving features in
BIOS on newer Xeon-D boards being shipped by Supermicro.

These are quite a few! I hope this helps someone :-)

Hyper-Threading ALL    [Enable]
Cores Enabled          [0 ALL]
EIST (P-States)        [Enable]
P State Domain         [ALL]
P-State Coordination   [HW_ALL]
Boot Performance Mode  [Max Performance]
Turbo Mode             [Enable]
Enable CPU HWPM        [Disable]
CPU C State            [Disable]
ACPI T-States          [Disable]
Energy Performance Tuning         [Disable]
Energy Performance BIAS Setting   [Performance]
Energy Efficient Turbo            [Disable]


On Fri, Dec 9, 2016 at 7:01 AM, Micky <[email protected]> wrote:

> Isn't it weird that I reduced the number of vCPUs to 2 and the issue
> hasn't occurred since?
>
>
> On Mon, Dec 5, 2016 at 9:37 PM, Robert Mustacchi <[email protected]> wrote:
>
>> On 12/4/16 17:29 , Micky wrote:
>> > Hey Robert, do you think it has something to do with HPET clock?
>> >
>> > I can see the Dtrace hpet probes on the machines for affected qemu
>> process.
>> > The probes work whether or not hpet is enabled or disabled, both in
>> machine
>> > BIOS or in Windows, via bcdedit /set useplatformclock.
>>
>> Most of the indicated threads suggest that they're all doing work in the
>> kernel, the user land process state isn't going to be very useful here.
>> I'd probably take a look at kvmstat to see what the guest is doing in
>> this state.
>>
>> Robert
>>
>> > On Sat, Dec 3, 2016 at 4:32 PM, Micky <[email protected]> wrote:
>> >
>> >> I'm noticing a Windows KVM crash without any logs or leaving any sort
>> of
>> >> trace.
>> >> It just freezes and VNC doesn't respond back.
>> >>
>> >> I'm on platform 20161101T004406Z and machine is a Xeon-D.
>> >>
>> >> It happened a couple of times last week but this time, I managed to
>> dump
>> >> the core with gcore.
>> >>
>> >> Full stack:
>> >> https://gist.github.com/anonymous/999ae4c371f9af4da7fe3eb40761af01
>> >>
>> >> Core dump:
>> >> http://158.69.218.12/core.10072.10072.gz
>> >>
>> >> Here's first thread in core stack. Thanks!
>> >>
>> >> # mdb -e '::status;$C' core.10072.10072
>> >> debugging core file of qemu-system-x86 (64-bit) from damned
>> >> file: /smartdc/bin/qemu-system-x86_64
>> >> initial argv:
>> >> /smartdc/bin/qemu-system-x86_64 -m 16384 -name
>> >> 1ba82163-a746-c28d-944a-813287cb
>> >> threading model: native threads
>> >> status: process core file generated with gcore(1)
>> >> fffffd7fffdf9110 libc.so.1`ioctl+0xa()
>> >> fffffd7fffdf9170 kvm_set_irq_level+0x65(f6b430, 0, 1, fffffd7fffdf91ec)
>> >> fffffd7fffdf91c0 kvm_set_irq+0x3d(0, 1, fffffd7fffdf91ec)
>> >> fffffd7fffdf9210 kvm_i8259_set_irq+0x31(101fe70, 0, 1)
>> >> fffffd7fffdf9250 qemu_set_irq+0x42(405438480, 1)
>> >> fffffd7fffdf9280 qemu_irq_pulse+0x21(405438480)
>> >> fffffd7fffdf92e0 update_irq+0x1cd(4011bc750, 1)
>> >> fffffd7fffdf9340 hpet_timer+0x1c4(4011bc750)
>> >> fffffd7fffdf9390 qemu_run_timers+0xb0(f63020)
>> >> fffffd7fffdf93a0 qemu_run_all_timers+0x51()
>> >> fffffd7fffdff430 main_loop_wait+0x480(0)
>> >> fffffd7fffdff460 kvm_main_loop+0x119()
>> >> fffffd7fffdff480 main_loop+0x17()
>> >> fffffd7fffdff7c0 main+0x2da4(32, fffffd7fffdff7e8, fffffd7fffdff980)
>> >> fffffd7fffdff7d0 _start+0x6c()
>> >>
>> >
>> >
>> 
>> 
>
>



-------------------------------------------
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125&id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Reply via email to