Some systems incorrectly reported as powered off / aborted

[kihjin]

VirtualBox 3.0.4
Host: Ubuntu Server 9.04

I have several VMs running in headless mode. When they were first started, their State was reported correctly.

For whatever reason, and this has happened frequently, the VirtualBox book keeper service forgets that they are running (only some of them). To check the state of all the systems, I use vm list vms --long | grep State. Currently, VirtualBox is reporting several systems as "powered off" or "aborted," when they are running just fine. They respond to ping, ssh and other network services.

These same systems do not appear in vm list runningvms

This is not good, because RDP stops working, and any attempt to communicate with the system using controlvm fails because as far as VirtualBox is concerned, the system is NOT running.

The first question is why does the service forget about them or lose connection with them? That is a difficult problem to solve, though should be solved nonetheless.

In the mean time, I'd like to force the service to rescan or reconnect with these faux-offline systems. How can this be accomplished? I'm guessing that if it is possible at all, it would require the use of the SDK. This is OK, I would just like to be pointed in the right direction.

[Perryg]

Very unusual, I think you should probably create a ticket on this one. Bugtracker link is below my post. You will need to create an account there though as it is on a different system. Please post the results here when you find out the problem so it can benefit anyone else that has this problem. Also if you could update your profile it would help the forum responders out.