VBoxHeadless segfaults when trying RDP with auth after VM has been running for a few days

[Michael]

I have a Fedora 10 64-bit server running 4 VMs in headless mode.

2 Xeon E5430 CPUs 16 GB RAM VirtualBox 2.2.2 64-bit

After the VMs have been running for more than a few days I try to connect to them via RDP. I have them all setup with RDP on separate ports with External auth. When I try to connect to any VM it fails and the VM dies. Here's the kernel output:

VBoxHeadless[3540]: segfault at 7f757e690400 ip 00007f759f484870 sp 00007f759f3d4f68 error 4 in libcrypto.so.0.9.8g.#prelink#.XColQA (deleted)[7f759f3d7000+140000]

If I try to connect to another VM with RDP it also dies. A simple "startvm" command gets things going again. I can use RDP again on the restarted VMs, but I'm afraid to use RDP now.

The two guest VMs I attempted RDP with were SCO Unix VMs. I have a Windows XP and Linux VM (for the grand total of 4 VMs) as well, but I do not want to try RDP on those as they need to stay running.

[Frank Mehnert]

The ​core dump would be probably interesting in that case...

[Michael]

I will attempt to retrieve a core dump, but I am not sure how long the VMs need to stay running before RDP is munged.

VirtualBox seems to be munged as well now as a VBoxManage controlvm "vm-name" poweroff command results in an error saying no such "vm-name" is registered.

[Michael]

Replying to mooninite:

VirtualBox seems to be munged as well now as a VBoxManage controlvm "vm-name" poweroff command results in an error saying no such "vm-name" is registered.

Ignore this. I was logged in as the wrong user.

[Michael]

Nothing to report... The machines have been up for over a week and I've been trying RDP at least once a day to make sure they don't crash. The only thing I can think of that caused this is that there was a package update run and it involved openssl. I think it's safe to close this as a non-VBox bug. Sorry for the spam.

[Sander van Leeuwen]

Thanks for the feedback.