CHR v7 on PVE 8.X (kernel 6.X) : one specific vNIC stops receiving traffic

auranext

Well-Known Member
Jun 5, 2018
55
2
48
124
Hello everyone,


On CHR v7 running on Proxmox, we are experiencing a recurring issue where, after an uncertain period of normal operation, one specific virtio vNIC stops receiving traffic.
  • Host side (Proxmox TAP interface):
    • Packet captures confirm that inbound traffic (IP packets and ARP requests from external sources) arrives correctly at the TAP.
    • However, no packets are delivered to the CHR guest.
    • log are fully OK, no error
  • Guest side (CHR):
    • Outbound ARP requests sent by the CHR are answered at the host level, but the replies never reach the CHR.
    • Packet captures inside the CHR show only the traffic generated locally by the CHR; no inbound frames are visible.
    • The RX counters on the affected ether interface stop increasing, confirming that inbound traffic is no longer delivered.


    • workaroud
  • Disabling and re-enabling the affected ether interface from within the CHR temporarily restores connectivity.
  • But... the issue reoccurs after an unpredictable amount of time.
  • It always affects the same vNIC.

  • Problematic setups:
    • CHR v7 on Proxmox VE 8.3.2 and 8.4.1 (kernel 6.x) → issue observed.
  • Working setups (no issue):
    • CHR v7 on Proxmox VE 5.3, 6.2, and 7.0.11 (kernel < 6) → no issue.

This leads to a complete loss of connectivity on the affected vNIC until manual intervention is performed.

Someone has notice this behavior ?
Thx for your help
 
Last edited: