Aborting Linux VM guests on VBox 7.2.6/7.2.8

[KloCZ2]

In a Using virtualbox / Discuss the VirtualBox 7.2.6 release here section I reported (Feb 11 2026) this :

Strange thing with 7.2.6 VBox:

* host (older Intel Xeon X5650, x86_64 level 2) is Fedora 42, kernel 6.18.X

* guest VM is Fedora 42/kernel 6.18 as well (guest is running some java applications with decent/average load).

When I upgraded host from 7.1.14 to 7.2.6, guest VM is aborting after 10-20 hours of run with this last line in VBox.log :

Code: Select all

Event: forcefully unregistering passive event listener 00007f42700014a0 due to excessive queue size

No matter, whether on guest are older 7.1.14 or newer 7.2.6 GuestAdditions, it's still aborting in the same way.

Downgrade to the VBox 7.1 line (the newest 7.1.16) solved this issue, it looks like (this guest is running several days without any problems).

Now I'm upgraded on the same machine VirtualBox to version 7.2.8 (host and guest are Fedora 43, with 7.0.4-100.fc43 kernel), situation remains the same, headless VM (just running some java applications) crashed/aborted after ~ 10-16 hours of run. In Vbox.log I can see the same Event, also I'm seeing this in kernel dmesg on host, could it be related :

[Wed May 13 18:51:29 2026] dxvk-cs[3804]: segfault at ffffffffffffffff ip 00007f9efacb1bc9 sp 00007f9e7e5d40d8 error 7 in libc.so.6[8ebc9,7f9efac93000+ff000] likely on CPU 21 (core 8, socket 1)
[Wed May 13 18:51:29 2026] Code: 00 00 00 48 83 fa 20 77 18 0f 11 07 0f 11 44 17 f0 c3 0f 11 47 e0 0f 11 47 f0 c3 66 0f 1f 44 00 00 48 3b 15 49 a6 15 00 77 5f <0f> 11 07 0f 11 47 10 48 01 d7 48 8
3 fa 40 76 d8 0f 11 40 20 0f 11

Now I'm reverting VirtualBox on host into the 7.1.18 version, I suppose everything will be fine there (as it was with 7.1.16), we will see...

[KloCZ2]

So, with version 7.2.12, I'm still observing the same situation (crash of VMs after several hours of run), even on a little bit newer server. My previous post described situation on old IBM x3650 M3 (processor Intel Xeon X5650), now I'm seeing the same situation on newer IBM x3650 M4 (processor Intel Xeon E5-2620 v2). What is interesting - on "just a small bit, if any" newer Dell Powerdge Rx20 series (processor Intel Xeon E5-2430 v2) VM is probably running fine (although I not heavily tested this there yet -- just I'm observing, that VM with no load is running there on VBox 7.2.12 third day without crash). That's why I'm thinking that this problem could depend on processor/hardware, even all the Xeons above are quite similar (both three have x86-64 v2 level, etc.).

About current tests with VBox 7.2.12 (host OS is AlmaLinux/RHEL 9, guest is Fedora 43) - the VM is crashing after several hours, no matter, whether there is some load inside VM or not - today I'm just start VM and don't run anything on it - so load was all the time 0.00 ... but it went to Aborted state in several hours anyway.

In the ~/VirtualBox VMs/<name_of_VM>/Logs/VBox.log I cannot see any significant message (last touch of this log was ca ~ 2 hours before the crash of VM).

Under the ~/.config/VirtualBox I can see in the VBoxSVC.log this :

Code: Select all

03:58:16.245738 DCon243  ERROR [COM]: aRC=VBOX_E_VM_ERROR (0x80bb0003) aIID={e36a5081-a82a-40bd-9e4e-42a44d6ce50f} aComponent={MachineWrap} aText={Failed to get a console object from the direct session (NS_ERROR_ABORT)}, preserve=false aResultDetail=0
03:58:16.245809 DCon246  ERROR [COM]: aRC=VBOX_E_VM_ERROR (0x80bb0003) aIID={e36a5081-a82a-40bd-9e4e-42a44d6ce50f} aComponent={MachineWrap} aText={Failed to get a console object from the direct session (NS_ERROR_ABORT)}, preserve=false aResultDetail=0
03:58:16.245858 DCon241  ERROR [COM]: aRC=VBOX_E_VM_ERROR (0x80bb0003) aIID={e36a5081-a82a-40bd-9e4e-42a44d6ce50f} aComponent={MachineWrap} aText={Failed to get a console object from the direct session (NS_ERROR_ABORT)}, preserve=false aResultDetail=0
03:58:16.245890 main     Saving settings file "~/VirtualBox VMs/<name_of_VM>/<name_of_VM>.vbox" with version "1.19-linux"
03:58:16.324891 Watcher  Reaper: Pid 17587 (0x44b3) was signalled: SIGSEGV (11 / 0xb)
03:58:16.346445 main     Finished saving settings file "~/VirtualBox VMs/<name_of_VM>/<name_of_VM>.vbox"

The time in log (03:58:16) corresponds to the length of run of that VM, so corresponds to the moment of the crash.

On all the three hosts/HWs VBox version 7.1.18 running fine / without any crash. But 7.1.x branch is unsupported today and doesn't work with newer updates of kernels/systems. So it's very unfortunate situation at all.