CentOS - Feb 2018 - Network broke for a while when libvirtd start

When start service libvirtd by

    service libvirtd start

the network will broke for a while and then reconnected. Following is
/var/log/messages:

Feb 24 13:28:51 node-0 kernel: lo: Dropping TSO features since no CSUM feature.
Feb 24 13:28:51 node-0 kernel: lo: Disabled Privacy Extensions
Feb 24 13:28:51 node-0 kernel: virbr0: Dropping TSO features since no CSUM
feature.
Feb 24 13:28:51 node-0 kernel: virbr0: Dropping TSO6 features since no CSUM
feature.
Feb 24 13:28:51 node-0 kernel: device virbr0-nic entered promiscuous mode
Feb 24 13:28:51 node-0 kernel: virbr0: starting userspace STP failed, starting
kernel STP
Feb 24 13:28:51 node-0 kernel: ip_tables: (C) 2000-2006 Netfilter Core Team
Feb 24 13:28:51 node-0 kernel: nf_conntrack version 0.5.0 (16384 buckets, 65536
max)
Feb 24 13:28:51 node-0 kernel: virbr0: topology change detected, propagating
Feb 24 13:28:51 node-0 kernel: virbr0: port 1(virbr0-nic) entering forwarding
state
Feb 24 13:28:51 node-0 kernel: 8021q: adding VLAN 0 to HW filter on device
virbr0
Feb 24 13:28:51 node-0 kernel: ixgbe 0000:02:00.0: eth2: detected SFP+: 3
Feb 24 13:28:52 node-0 kernel: ixgbe 0000:02:00.1: eth3: detected SFP+: 4
Feb 24 13:28:52 node-0 kernel: bonding: bond0: link status definitely down for
interface eth2, disabling it
Feb 24 13:28:52 node-0 kernel: virbr0: port 1(virbr0-nic) entering disabled
state
Feb 24 13:28:52 node-0 kernel: bonding: bond0: link status definitely down for
interface eth3, disabling it
Feb 24 13:28:53 node-0 kernel: ixgbe 0000:02:00.0: eth2: NIC Link is Up 10 Gbps,
Flow Control: RX/TX
Feb 24 13:28:53 node-0 kernel: bond0: link status definitely up for interface
eth2, 10000 Mbps full duplex.
Feb 24 13:28:53 node-0 kernel: ixgbe 0000:02:00.1: eth3: NIC Link is Up 10 Gbps,
Flow Control: RX/TX
Feb 24 13:28:53 node-0 kernel: bond0: link status definitely up for interface
eth3, 10000 Mbps full duplex.

OS: CentOS 6.5
libvirt: libvirt-daemon-1.2.5-1.mira1.x86_64
Does anyone know the reason?

Hello
> When start service libvirtd by
>
>     service libvirtd start
>
> OS: CentOS 6.5
> libvirt: libvirt-daemon-1.2.5-1.mira1.x86_64
https://www.redhat.com/mailman/listinfo/libvirt-users
is the official mailing list for the libvirt daemon, maybe you can ask
in there and get better support.
> Does anyone know the reason?
This mailing list is for the general usage of the CentOS system, I am
afraid that not many experts about libvert could be found here.

If you ask in the most appropriate mailing list, you may get the
quickest and the most helpful assistance.

Never mind, I have found the reason. When libvirtd started, it will create
bridge virbr0 and then write 1 to /proc/sys/net/ipv4/ip_forward (0 in default).
Set ip_forward from 0 to 1 will cause ixgbe driver reset NIC, so the network
broke for a while.

> -----Original Messages-----
> From: "Genghuang Wang" <wangtianjiao.wang959 at gmail.com>
> Sent Time: 2018-02-24 14:34:16 (Saturday)
> To: "CentOS mailing list" <centos at centos.org>
> Cc: 
> Subject: Re: [CentOS] Network broke for a while when libvirtd start
> 
> Hello
> 
> > When start service libvirtd by
> >
> >     service libvirtd start
> >
> 
> > OS: CentOS 6.5
> > libvirt: libvirt-daemon-1.2.5-1.mira1.x86_64
> 
> https://www.redhat.com/mailman/listinfo/libvirt-users
> is the official mailing list for the libvirt daemon, maybe you can ask
> in there and get better support.
> 
> > Does anyone know the reason?
> 
> This mailing list is for the general usage of the CentOS system, I am
> afraid that not many experts about libvert could be found here.
> 
> If you ask in the most appropriate mailing list, you may get the
> quickest and the most helpful assistance.
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> https://lists.centos.org/mailman/listinfo/centos