Solaris 10 host, Ubuntu 10.04 Guest, VBoxHeadless hangs

[srv1973]

Hi all,

I am running VBox 3.2.8 on Solaris 10 (Sun x4275 server) in Headless mode, with an Ubuntu 10.04 guest (32bit, installed from ubuntu-10.4-desktop-i386.iso, guest additions installed). I am starting this VM from an /etc/init.d script as:

nohup VBoxHeadless --startvm vm1 &

After that I access either via RDP or via SSH. This works fine for a while, but I've noticed that after a period of inactivity, the VM seems to hang. I'm not really sure how long it takes to 'hang', but if I start it in the afternoon, use it, go home and come back the next morning, I have trouble accessing the VM. The symptoms are:

• If I try to access via RDP I get the VM screen, but it's not responding
• If I try to access via SSH, I get the connection established, sometimes I see the MOTD, but I never get the shell prompt
• Stopping the VM via "VBoxManage controlvm vm1 acpipowerbutton" doesn't work (I usually use 'kill' to get rid of the VBox processes)

Has anybody seem similar behaviour before? Is there anything I can do to prevent this hang? To recover from this hang? To debug this?

Rgds, Maarten

[Perryg]

This could be a tough one if it is not as simple as making sure that the guest screen saver and power management are disabled.

[gu99roax]

I have experienced this problem with other guest OSes as well. What I did was logging in to the server via ssh to start a virtual machine via nohup. As soon as I logged out the machine hung. When using screen instead I don't have this problem.

[zebra]

Unfortunately, I can also confirm this one as to the following:

The Linux Ubuntu 10.04 LTS/Server guest seems to lock up only under high I/O (5MB/sec, ha!) scenarios, or, alternatively as another explanation, when CPU time seem fairly high/non idle for a significant period of time. Am happy to invest some time into helping you guys fix this, if there is anyone out there in VirtualBox team land that wants to help...

Details:

Host OS:

Code: Select all

zebra@rapoosev5:~$ uname -a
SunOS rapoosev5 5.11 snv_134 i86pc i386 i86pc Solaris

VBox Interface:

Code: Select all

zebra@rapoosev5:~$ VBoxHeadless -startvm UbuntuServer
Oracle VM VirtualBox Headless Interface 3.2.8
(C) 2008-2010 Oracle Corporation
All rights reserved.

Guest OS:

Code: Select all

root@sabhost:~# uname -a
Linux sabhost 2.6.32-24-server #43-Ubuntu SMP Thu Sep 16 16:05:42 UTC 2010 x86_64 GNU/Linux

Host system details:

Code: Select all

zebra@rapoosev5:~$ prtdiag
System Configuration: Gigabyte Technology Co., Ltd. P35-DQ6
BIOS Configuration: Award Software International, Inc. F9B 09/17/2008

==== Processor Sockets ====================================

Version                          Location Tag
-------------------------------- --------------------------
Intel(R) Core(TM)2 Quad CPU      Socket 775

==== Memory Device Sockets ================================

Type        Status Set Device Locator      Bank Locator
----------- ------ --- ------------------- ----------------
unknown     in use 0   A0                  Bank0/1
unknown     in use 0   A1                  Bank2/3
unknown     in use 0   A2                  Bank4/5
unknown     in use 0   A3                  Bank6/7

==== On-Board Devices =====================================

==== Upgradeable Slots ====================================

ID  Status    Type             Description
--- --------- ---------------- ----------------------------
0   available PCI              PCI
1   available PCI              PCI

Guest OS Details/arch/procinfo:

Code: Select all

root@sabhost:~# cat /proc/cpuinfo 
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 15
model name	: Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz
stepping	: 11
cpu MHz		: 2187.243
cache size	: 6144 KB
physical id	: 0
siblings	: 3
core id		: 0
cpu cores	: 3
apicid		: 0
initial apicid	: 0
fpu		: yes
fpu_exception	: yes
cpuid level	: 5
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx lm constant_tsc rep_good pni ssse3 lahf_lm
bogomips	: 4374.48
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 1
vendor_id	: GenuineIntel
cpu family	: 6
model		: 15
model name	: Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz
stepping	: 11
cpu MHz		: 2187.243
cache size	: 6144 KB
physical id	: 0
siblings	: 3
core id		: 1
cpu cores	: 3
apicid		: 1
initial apicid	: 1
fpu		: yes
fpu_exception	: yes
cpuid level	: 5
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx lm constant_tsc rep_good pni ssse3 lahf_lm
bogomips	: 4324.36
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 2
vendor_id	: GenuineIntel
cpu family	: 6
model		: 15
model name	: Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz
stepping	: 11
cpu MHz		: 2187.243
cache size	: 6144 KB
physical id	: 0
siblings	: 3
core id		: 2
cpu cores	: 3
apicid		: 2
initial apicid	: 2
fpu		: yes
fpu_exception	: yes
cpuid level	: 5
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx lm constant_tsc rep_good pni ssse3 lahf_lm
bogomips	: 4354.83
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

[srv1973]

Hi, I'm the original poster...

Strange thing is, after I posted this I restarted the VBox instance (for about the 100th time). Ever since, I have had no issues. There is one difference with how I used the VBox before:

I used to always use the RDP server to control the Guest OS. This time, I've not touched RDP and have only accessed the VM using SSH (and X over SSH, or rather I use x2go to get a graphical desktop).

Could it be possible that the built-in RDP server is somehow to blame? Do the other people who reported this issue use RDP?

Rgds

[zebra]

Possible, I suppose - but I've never actually used the V-RDP functionality of the host/client interface. This box is pure SSH. That isn't to say that the VRDP isn't turned on already.

I will turn it off, and see if the behaviour changes. I find it unlikely however. My guess is there is something more complicated happening I/O subsystem wise, as the behaviour seems to be triggered (for me at least) in high IO/CPU Utilisation scenarios...

z

[zebra]

.

Isolated. It's SMP related. Removed a vCore and made it a non SMP host, effectively uni-processor, and it's solid as a rock/stable.

Logging this as a bug.

z