[SOLVED] Proxmox 4.2 KVM Bridge packet loss

N

nade

New Member
Aug 26, 2015
16
1
3
Hi,

I just recognized that i ahve massiv package lost problems with my newly configured proxmox host.

PVE version: 4.2-11/2c626aa1
hoster: hetzner
Kernel: Linux 4.4.8-1-pve

I just setup check_mk for my monitoring and just recognized that i have a fluctuating packet loss of 2 - 35% on my kvm vm´s.

Test VM 1: CRIT - [eth0] (up) speed unknown, in: 96.33 B/s, in-errors: 28.10%CRIT >= 0.1%, out: 344.40 B/s
Test VM 2: CRIT - [eth0] (up) speed unknown, in: 66.03 B/s, in-errors: 26.44%CRIT >= 0.1%, out: 435.34 B/s

The difference between the packetloss between the Vm´s is very small, (1-4%)

This is my network config:

Code: 
### Hetzner Online GmbH - installimage
# Loopback device:
auto lo
iface lo inet loopback
iface lo inet6 loopback

# device: eth0
auto  eth0
iface eth0 inet static
  address   78.46.xxx.184
  netmask   255.255.255.224
  gateway   78.46.xxx.161
  # default route to access subnet
#  up route add -net 78.46.xxx.160 netmask 255.255.255.224 gw 78.46.xxx.161 eth0

iface eth0 inet6 static
  address 2a01:4f8:xxx:1270::2
  netmask 64
  gateway fe80::1

auto vmbr0
iface vmbr0 inet static
    address  178.63.xxx.193
    netmask  255.255.255.240
    bridge_ports none
    bridge_stp off
    bridge_fd 0

    #subnet
    up ip route add 178.63.xxx.194/28 dev vmbr0
    up ip route add 178.63.xxx.195/28 dev vmbr0
    up ip route add 178.63.xxx.196/28 dev vmbr0
    up ip route add 178.63.xxx.197/28 dev vmbr0
    up ip route add 178.63.xxx.198/28 dev vmbr0
    up ip route add 178.63.xxx.199/28 dev vmbr0
    up ip route add 178.63.xxx.200/28 dev vmbr0
    up ip route add 178.63.xxx.201/28 dev vmbr0
    up ip route add 178.63.xxx.202/28 dev vmbr0
    up ip route add 178.63.xxx.203/28 dev vmbr0
    up ip route add 178.63.xxx.204/28 dev vmbr0
    up ip route add 178.63.xxx.205/28 dev vmbr0
    up ip route add 178.63.xxx.206/28 dev vmbr0

Code: 
# The loopback network interface
auto lo
iface lo inet loopback

# The primary network interface
allow-hotplug eth0
iface eth0 inet static
        address 178.63.xxx.195
        netmask 255.255.255.240
        network 178.63.xxx.192
        broadcast 178.63.xxx.207
        gateway 178.63.xxx.193
        # dns-* options are implemented by the resolvconf package, if installed
        dns-nameservers 8.8.8.8 8.8.4.4

traceroute:

Code: 
guest@dnstools.ch:~> traceroute 78.46.xxx.184
1 gw.f.netclusive.de (89.110.131.1) 1.204 ms
2 89.202.113.81 (89.202.113.81) 5.720 ms
3 te2-1-111.bg1.fr2.eu.equinix.net (217.68.155.165) 0.286 ms
4 decix2-gw.hetzner.de (80.81.193.164) 10.598 ms
5 core21.hetzner.de (213.239.245.13) 5.156 ms
6 juniper2.rz11.hetzner.de (213.239.245.210) 4.976 ms
7 hos-tr4.ex3k1.rz11.hetzner.de (213.239.228.98) 5.781 ms
8 *
9 *
10 *

Code: 
guest@dnstools.ch:~> traceroute 178.63.xxx.195
1 gw.f.netclusive.de (89.110.131.1) 0.693 ms
2 89.202.113.81 (89.202.113.81) 0.986 ms
3 te2-1-111.bg1.fr2.eu.equinix.net (217.68.155.165) 0.306 ms
4 decix2-gw.hetzner.de (80.81.193.164) 0.353 ms
5 core22.hetzner.de (213.239.245.17) 5.125 ms
6 juniper2.rz11.hetzner.de (213.239.245.114) 4.955 ms
7 hos-tr3.ex3k1.rz11.hetzner.de (213.239.228.66) 5.713 ms
8 censored.de (78.46.xxx.184) 5.236 ms
9 *
10 *
11 *

With same config on PVE 4.0 i have no packet loss.

anyone have an Idea?
 
Ok, i Maybe Solved the Problem by myselve.

I Just changed the Network Model in Hardware Settings from Intel E1000 to VirtIO.

Will update later if my packetl oss is away for a bit longer now.
 
Ok, so now for one hour 0% packetloss, so the solution is to change the network model to VirtIO.
For Linux distribtions this is no problem cuase VirtIO drivers are included.

For Windows it might be a problem.

Regards
 
You must log in or register to reply here.
Top Bottom