[PVE-User] packet drop on vmbr0

DEVARIEUX Alain alain.devarieux at sib.fr
Thu Jun 15 11:46:04 CEST 2017

The promox version is 4.4-1 not 2.3.4 as I said.

-----Message d'origine-----
De : pve-user [mailto:pve-user-bounces at pve.proxmox.com] De la part de DEVARIEUX Alain
Envoyé : Thursday, June 15, 2017 11:12 AM
À : pve-user at pve.proxmox.com
Objet : [PVE-User] packet drop on vmbr0

Hi all,

We are experiencing an issue on 2 proxmox nodes. They are newly installed 2.3.4 servers. One is master and syncing with pfsync to the second one. Sync works well. Each of these proxmox nodes hosts 2 VMs : a pfsense for firewalling and a vyos for bgp sessions with our ISP.
We have an OpenVswitch bridge vmbr0 linked to eth4, a 10Gb physical interface. (Broadcom NetXtreme II, BCM57810).

I have no errors on the physical interface, but on vmbr0 I can see nearly 10% dropped packets. (on both proxmox nodes)
vmbr0     Link encap:Ethernet  HWaddr 00:0e:1e:7c:68:80
          inet6 addr: fe80::20e:1eff:fe7c:6880/64 Scope:Link
          RX packets:8154568 errors:0 dropped:1136061 overruns:0 frame:0
          TX packets:157423 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1
          RX bytes:1631993883 (1.5 GiB)  TX bytes:120943834 (115.3 MiB)

The VMs are using e1000 drivers.  I saw on several topics that switching to virtIO seems to solve this dropped packets problem. On the other way, CARP VIP documentation for pfsense advise to use e1000 drivers or CARP may not work. https://doc.pfsense.org/index.php/Configuring_pfSense_Hardware_Redundancy_(CARP)#KVM.2BQEMU_Users

Am I in a deadlock issue?
What would you advise me to do?

Alain Devarieux
Pôle Infrastructure

pve-user mailing list
pve-user at pve.proxmox.com

More information about the pve-user mailing list