[SOLVED] Flapping network interface in KVM-Box for Proxmox 4.2

paderepiktet

New Member
Jun 13, 2016
11
2
3
60
Hello,

first of all, thank you for providing this great software to the community! During the last year we have made great experiences with two Proxmox-Servers of version 3.4.

i) Now a Proxmox-Version 4.2 has been installed on a fresh hardware at the same hoster
with just a slight change in the network configuration. There are three KVM-Boxes (originally coming from one Proxmox 3.4-Server by backup and restore) and each of these Boxes has three IP-adresses assigned to it (static eth0, eth0:0 and eth0:1).

ii) The Boxes are connected to the sourrounding net via "Briged mode". The boxes seem to work well and are operational and the infrastructure provider has approved the IP-configuration of the boxes.

iii) The KVM - Box has the following settings:

VLAN Tag:
Bridge: vmbr0
Firewall: on
Model: Intel E1000,
see [2] for the output of qm config.

iv) But the boxes are monitored by check_mk via the preconfigured interface-check lnx_if. This check sends packages to the interface from within the KVM-Box and reports in-errors:

CRIT - [eth0] (up) MAC: 62:62:32:65:32:66, 1 Gbit/s, in: 12.03 kB/s, in-errors: 0.19%(!!) = 0.1, out: 845.11 B/s

Performance Data: rta=0.301ms;200.000;500.000;0; pl=0%;40;80;; rtmax=0.488ms;;;; rtmin=0.237ms;;;;
Performance Data: in=12316.117018;;;0;125000000 inucast=195.222016;;;; innucast=0;;;; indisc=0;;;; inerr=0.366239;0.01;0.1;; out=845.113062;;;0;125000000 outucast=1.398367;;;; outnucast=0;;;; outdisc=0;;;; outerr=0;0.01;0.1;; outqlen=0;;;0;

For the output of ethtool, that if used by lnx_if, see [1].

v) According to the in-errors: 0.19%(!!) in-errors: 0.19%(!!) = 0.1 the monitoring server reports critical errors. The interface is reported to be flapping.

vi) This leads to the following questions:

- What are the best steps to pin down this error message to the actual cause? Which files, tools shall be applied?

- Which configuration parameters in proxmox typically influence this?

- Do we have to change something in the monitoring or in proxmox or in both?

The strange point is that check_mk and Proxmox 3.4 do not show such error messages.

Any help is highly appreciated!

With best regards

Andreas


[1]: Output of ethtool eth0:

ethtool eth0
Settings for eth0:
Supported ports: [ TP ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Supported pause frame use: No
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Advertised pause frame use: No
Advertised auto-negotiation: Yes
Speed: 1000Mb/s
Duplex: Full
Port: Twisted Pair
PHYAD: 0
Transceiver: internal
Auto-negotiation: on
MDI-X: off (auto)
Cannot get wake-on-lan settings: Operation not permitted
Current message level: 0x00000007 (7)
drv probe link
Link detected: yes

[2]: Output of qm config:

balloon: 4096
bootdisk: ide0
cores: 6
description: ...
ide0: local-lvm:vm-105-disk-1,size=1002G
ide2: none,media=cdrom
memory: 32768
name: real.shk-datacheck.de
net0: bridge=vmbr0,firewall=1,e1000=66:61:36:36:37:37
numa: 0
ostype: l26
smbios1: uuid=e9853546-39d2-4fae-827b-814b6354a996
sockets: 2
 
Sorry for the necro post, but it might help others: We changed to vmxnet3 instead of e1000. Hopefully it will stay stable (stable for 4 days now). Sure, not the performance, but enough for our use case: Migration of an old XEN host with nested virtualization to new hardware. The App running in NetBSD is needing a special version of Xen (4.1.4) to run.

virtio is not possible on nested virtualization (nested host does not boot).
 
I have the same problem.
Vhost i a Windows 10 LTSC version.
I try VirtIO and vmxnet3. I change the switch, cables, router, PC. and all that i remember
:(
Some solution alternative?