VBoxHeadless talks to the X Server?! / fatal IO error 113

[tbroberg]

I need to be able to start a bevy of VMs remotely, and the method I thought would work is crashing VMs.

I am cloning a test network with 5 separate machines on one virtual host.

The host is in the office, and often I am working from home, so I would like to be able to start up headless and control the VMs remotely.

Rather than do this from a script at boot-time, I'm firing them up manually with a script run from an ssh session (typically from within a screen session.) Typical syntax:

VBoxHeadless -s myvm -p 3890&

The catch is that if I'm logged in from the vpn, the VMs eventually appear to crash. On my screen session I get an error message about an IO error trying to talk to the X server that includes the string 'virtualbox "fatal IO error 113 (No route to host)"' and the address of my home system assigned by DHCP the last time I logged into the VPN.

In the most recent case, it complained that it could not reach 192.168.2.113 when the current address is 192.168.2.116.

Now the questions:
1 - Why should VBoxHeadless be trying to talk to the X Server?
2 - Any recommendations for a safe way to start up headless sessions to bypass this issue? How does everybody else do this?

TIA,
- Tim.

[TerryE]

I notice that you are not redirecting stdout and stderr. I would have expected something more like:

• nohup VBoxHeadless -s $VM -p $PORT >> $LOG_DIR/$VM.$PORT.log 2>&1 &

where variables are set in your startup script (or redirect to /dev/null if you are going to check the VMs status by doing a VBoxManage command).

[tbroberg]

I notice that you are not redirecting stdout and stderr. I would have expected something more like...

Ah! I'll try that and see if it helps.

Thanks,
- Tim.