Performance issues

[Maximus80]

Hi All!

I'm at a loss regarding the following topic

I migrated 3 Fedora vms from an old Fedora based server to a new Fedora based server (As I know export/import is not the recommended method for moving a vms, so I manually copied the vms to the new server and added them into VirtualBox). Guest Additions are also upgraded on the vms.

These are biggere vms, 700 GB and 1700 GB, both use 32 GB RAM and use 4 and 10 CPU cores.

When I start these vms, I experience performace problems, hovever there are zero load (load average is about 0). When I ping these machines from same subnet, the 60% of packets are lost. The characters entered into the terminal are lost or repeated. So it seems that the both OS run are constantly stalling/interrupting in the new server.
Outgoing ping from these vms also bad, sometimes one or more ping lines are missed, as if waiting a bit.

The old server is running with Fedora 32 and VirtualBox 6.1.10
The new one is running with Fedora 41 and VirtualBox 7.1.6.
These are strong servers, the new is a DELL r750xs.

I cannot imagine why these vms don't run perfectly. They have plenty of resources.

There is a smaller migrated vms which running fine. Pinging fine, typing fine!

Is there any known bug in the VirtualBox 7.1.6. about performance?
I found an issue about allocating memory (viewtopic.php?t=112438), I run into this problem yesterday, but when I execute the recomended command (echo 3 | sudo tee /proc/sys/vm/drop_caches) the machines start. But maybe there are other bug connected to RAM and performance.

[Maximus80]

I found the solution.
I had to regenerate the MAC addresses on the new host's VirtualBox.
After the vms have new MAC address, the network issues have disappeared.

[klaus]

Did you have both the old and new VMs (which must have been using bridged networking, something you didn't state) active in the same LAN? That creates a "duplicate MAC" issue which can cause all kinds of strange network connectivity issues for both.

[Maximus80]

Yes, the old and new VMs use bridged network and they were active in the same LAN in the same time, howeveer they have different ip addresses. We don't use nothing with MAC (like MAC filtering) but maybe our network devices use MAC address in their tables (Ethernet switches keeps a table of which MAC addresses are attached to which ports...)

[klaus]

Exactly. It's the switches which sabotage you in such duplicate MAC cases even if you manually made sure that the IP address is different.