[ovirt-users] Bizzare oVirt network problem

Pavel Gashev Pax at acronis.com
Thu Jul 13 06:15:06 UTC 2017


Fernando,

The issue can be triggered by a packet with the same source MAC address as VM has. If it’s received by network equipment from some different port, your VM could stop receiving network traffic.

There are two options. Either there is a network topology issue, or there is another VM with the same MAC address. The last is possible if you have another oVirt installation (testing? staging?) in the same network segment.


From: Fernando Frediani <fernando.frediani at upx.com.br> on behalf of FERNANDO FREDIANI <fernando.frediani at upx.com>
Date: Wednesday, 12 July 2017 at 23:10
To: Pavel Gashev <Pax at acronis.com>, "users at ovirt.org" <users at ovirt.org>
Subject: Re: [ovirt-users] Bizzare oVirt network problem


Hello Pavel

What you mean by another oVirt instance ? In one Datacenter it has 2 different clusters (or Datacenter in oVirt way of orrganizing things), but in the other Datacenter the oVirt Node is standlone.

Let me know.

Fernando

On 12/07/2017 16:49, Pavel Gashev wrote:
Fernando,

It looks like you have another oVirt instance in the same network segment(s). Don’t you?


From: <users-bounces at ovirt.org><mailto:users-bounces at ovirt.org> on behalf of FERNANDO FREDIANI <fernando.frediani at upx.com><mailto:fernando.frediani at upx.com>
Date: Wednesday, 12 July 2017 at 16:21
To: "users at ovirt.org"<mailto:users at ovirt.org> <users at ovirt.org><mailto:users at ovirt.org>
Subject: [ovirt-users] Bizzare oVirt network problem

Hello.

I am facing a pretty bizzare problem in two of my Nodes running oVirt. A given VM running a few hundred Mbps of traffic simply stops passing traffic and only recovers after a reboot. Checking the bridge with 'brctl showmacs BRIDGE' I see the VM's MAC address missing during this event.

It seems the bridge simply unlearn the VM's mac address which only returns when the VM is rebooted.
This problems happened in two different Nodes running in different hardware, in different datacenter, in different network architecture, different switch vendors and different bonding modes.

The main differences these Nodes have compared to others I have and which don't show this problem are:
    - The CentOS 7 installed is a Minimal installation instead of oVirt-NG
    - The Kernel used is 4.12 (elrepo) instead of the default 3.10
    - The ovirtmgmt network is used also for the Virtual Machine showing this problem.

Has anyone have any idea if it may have anything to do with oVirt (any filters) or any of the components different from a oVirt-NG installation ?

Thanks
Fernando




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20170713/c9e87836/attachment.html>


More information about the Users mailing list