Random kvm network drop with SolusVM + OVH

icekev

Hello Folks,

I bought a dedicated server from OVH and have been trying to make things right.
OpenVZ worked like a charm, took one of OVH solusvm slave OVZ template and never had any issues with that.

But time passes and people grow, so It's time to use KVM - Right.

For the purpose of this post :
Main server IP : 1.1.1.1
16 ip subnet : 2.2.2.2
Operating system : CentOS 6.6

I set-up a KVM slave, partitionned my drives like mentionned on SolusVM site.

80GB ext4 /
4GB swap
xGB ext4 lv /vps

I somewhat configured the bridge properly, because yes my VPSes are connecting perfectly to the internet.

Here is my configuration :

/etc/sysconfig/network-scripts/ifcfg-eth0
    DEVICE=eth0
    ONBOOT=yes
    BRIDGE=br0


/etc/sysconfig/network-scripts/ifcfg-br0
    DEVICE=br0
    TYPE=Bridge
    BOOTPROTO=static
    IPADDR=1.1.1.21
    NETMASK=255.255.255.0
    ONBOOT=yes
    GATEWAY=1.1.1.254
    DELAY=0
    IPV6INIT=yes
    IPV6_AUTOCONF=no
    IPV6ADDR=xxxx:xxxx:xxx:xxxx::/64
    ARPCHECK=no


/etc/sysconfig/network-scripts/route-eth0
    1.1.1.254 dev eth0
    default via 1.1.1.254 dev eth0


Enabled forwarding of IPv4 in 
    /etc/sysctl.conf
    net.ipv4.ip_forward = 1


Flushed IP tables 
    iptables --flush
    service iptables restart

And then on SolusVM, configured my IP block to look like this :

Gateway : 2.2.2.142   (IP's range from 128-143)
Netmask: 255.255.255.240
ns1/2 : 8.8.8.8/8.8.4.4

Now for some reasons, all work super fine. Network is good, VM are all chill. I created like 5 VM to test this thing out.

If i let it be for a few days, sometimes after a random amount of time (Vary of 4hours to a few days) 4 out of 5 VM will lose network connectivity. It wont come back automatically, you need to either Reconfigure network, or reboot the machine. Sometimes none of these works too.

Anybody has any experience trying to setup KVM with SolusVM on a CentOs OVH dedicated server ?

Extra : The VM's still run fine after network drop. You can log into it using VNC and nothing particular is going on. It simply cannot connect to the internet anymore.

Extra2: Upon network drop, the main server is still reachable. Nothing is wrong with main server, it only seems to affect the VMs

Is this more of a SolusVM issue or something to do with my network configuration ?

Any help would be appreciated

Resources used :
https://www.lowendtalk.com/discussion/32361/installing-configuring-kvm-on-ovh-with-solusvm
http://help.ovh.ie/BridgeClient

MikePT

Hmmm, are you assigning OVH's IP Addressec Mac Addresses?

If not, that's probably why.

icekev

Oh yeah i forgot to mention : I went into the ovh manager and generated Vmac and assigned them on each of my IP manually instead of the default "Auto" thing in solusvm

agonyzt

I've had a similar issue in the past where OVH would randomly "blacklist" my failover IPs when they would send too many ARP requests. It had something to do with the VMs exposing their virtual MAC on the network. Proxying the ARP requests on br0 fixed my issue, but I didn't have a whole subnet, just a few failover IPs. Might be totally different for you!

icekev

They might be failover IPs who know, i just ordered 16 IPs.
How did you do that network proxy thing in the Br0 ?

Not too sure to understand the meaning of that :P