Network traffic drops on node machines

[gokhanm]

Hi all,

We use Proxmox VM 8.3.0 cluster with 2 node machine. We have 10Gb ethernet card and 1Gb ethernet card on 2 node machine. We create linux bridge over 10gb ethernet card.

Every node machines have network traffic drops. We checked whether there was any packet drop on the switches. There is no drop visible on the switches .
Thinking that traffic drops might be caused by an error on the Linux bridge, we set up two machines to communicate via 1Gb ports. We configured the VM servers to access the internet via a 10Gb Linux bridge. However, this did not resolve the drops.
Could this be due to a configuration error on Proxmox? We can't understand why there is a traffic drop.


Our 10GB ethernet card model is Mellanox Technologies MT27710 Family [ConnectX-4 Lx].

1 GB ethernet card models;
Broadcom Inc. and subsidiaries NetXtreme BCM5719 Gigabit Ethernet PCIe (rev 01)

 

[sinanc]

Hi everyone,


I’m experiencing the same issue on my setup.
I have a Proxmox VE 8.3.0 cluster with two nodes, both using the hardware listed below.
Just like the original post mentioned, I also moved my cluster network to 1 Gbps adapters to reduce packet drops.


In my case, I managed to fix the packet drop issue inside the VMs by switching the virtual NIC model to VMware vmxnet3, but I’m still seeing packet drops on the physical nodes themselves, and I haven’t been able to solve that yet.


Has anyone else faced the same issue?
If so, could you please share what the root cause or solution was?

---

Hardware Information​

Server model: HPE ProLiant DL380 Gen11
BIOS version: 2.44 (01/17/2025)
Firmware revision: 1.66
CPU: 2× Intel Xeon Gold 6530 @ 2.1 GHz (128 threads total)
Network adapters:

• Mellanox ConnectX-4 Lx (10/25 GbE) — mlx5_core driver, firmware 14.23.8036 (HP_2420110034)
• Broadcom NetXtreme BCM5719 Gigabit Ethernet — tg3 driver, firmware v1.55 NCSI v1.5.55.0

---

Thanks in advance for any help or insights!


Best regards,