Random crash VM and Nat Network together

[box-tester]

Hello!

Host Ubuntu Server 17.10 Have here Nat network and 3 VMs: 1) Win2016 2) Win2016 (Exchange) 3) Debian 9 (openvpn)

Random crash 1-5 days with VM №2 (Exchange) with blackscreen and Nat network on other VMs has not ping.

Poweroff all VMs (Exchange with emergency stop) and restart virtualbox service help for anothe 1-5 day.

[box-tester]

crash vm log

[Valery Ushakov]

Is the VBoxNetNAT process still running when that happens? Anything interesting in the natnet log, in your case it should be cod-tdg.log in either ~/.config/VirtualBox or ~/.VirtualBox?

[box-tester]

Replying to vushakov:

Is the VBoxNetNAT process still running when that happens? Anything interesting in the natnet log, in your case it should be cod-tdg.log in either ~/.config/VirtualBox or ~/.VirtualBox?

yes, it is. cod-tdg.log has nothing interesting - no errors After crash, i tested natnetwork: 1) VM crashed - service VBoxNetNat is on 2) another VM in ths natnetwork is work and ping sites. wget works one time 3) start crashed vm. wget do not work. ping is work (from inside to outside). natnetwork work unstable. 4) stop all vm's and restart /etc/init.d/virtualbox 5) start 1 vm with debian: wget is work 6) start other VM's - all is work correctly

[box-tester]

i'm check i/o cache on all vms, power on KVM paravirtualization on VM-1, add user to group vboxusers and restart all VMs. Waiting...

[box-tester]

Another crash: vboxnetnat. VM 2 is online. On all VMs choose Paravirtualization: Default. Restart Vms and virtualbox service.

[box-tester]

Another crash: VBoxNetNat. All VM is online. Restart virtualbox service.

[Valery Ushakov]

Can you obtain a stack trace from the crash?

[box-tester]

OVH kernel installed and two weeks uptime and all it's ok! OVH has some bugs with original kernels.

Please, close this ticket.