Hi All,
I see an unexpected beahviour from NetworkManager on CentOS 7.1. Using nmcli tool, I create a bond with two slaves as explained in the Red Hat 7.1 Networking guide. I enable slaves and master; bond works as expected. When I restart NetworkManager, it creates a new bond with the same name but not connected to any device. Two bonds with the same name is confusing for my other monitoring scripts. I'm wondering why a second bond is created? Is it a bug in NetworkManger?
#Create a bond with two slaves nmcli con add autoconnect no type bond con-name bond0 ifname bond0 nmcli con mod bond0 ipv6.method ignore ipv4.method manual ipv4.addresses ${BOND_IP}/${BOND_CIDR} ${BOND_GW} ${BOND_DNS} ${BOND_DNS_SEARCH} ipv4.never-default no ipv4.ignore-auto-dns no nmcli con add autoconnect no type bond-slave con-name bond-slave-eth0 ifname eth0 master bond0 nmcli con add autoconnect no type bond-slave con-name bond-slave-eth1 ifname eth1 master bond0
#Enable bond nmcli con mod bond-slave-eth0 connection.autoconnect yes nmcli con up bond-slave-eth0
nmcli con mod bond-slave-eth1 connection.autoconnect yes nmcli con up bond-slave-eth1
nmcli con mod bond0 connection.autoconnect yes nmcli con up bond0
systemctl restart NetworkManager systemctl restart iptables
nmcli con | grep bond bond0 9942bdc6-df72-4723-b2ed-47a78e3a5c59 bond bond0 bond-slave-eth0 8b0fbbe1-a7f0-448c-8005-46d11599f57a 802-3-ethernet eth0 bond-slave-eth1 333dd1b9-15a4-4119-8e42-55ac3621a85d 802-3-ethernet eth1 *bond0 460dd9e8-bc0b-473e-9c89-41facda98b66 bond -- # Why this extra bond connections has been created?*
I'd appreciate your comments and suggestions to fix the issue.
Thanks,
Joe
On 07/07/16 05:21 PM, Joe Smithian wrote:
Hi All,
I see an unexpected beahviour from NetworkManager on CentOS 7.1. Using nmcli tool, I create a bond with two slaves as explained in the Red Hat 7.1 Networking guide. I enable slaves and master; bond works as expected. When I restart NetworkManager, it creates a new bond with the same name but not connected to any device. Two bonds with the same name is confusing for my other monitoring scripts. I'm wondering why a second bond is created? Is it a bug in NetworkManger?
#Create a bond with two slaves nmcli con add autoconnect no type bond con-name bond0 ifname bond0 nmcli con mod bond0 ipv6.method ignore ipv4.method manual ipv4.addresses ${BOND_IP}/${BOND_CIDR} ${BOND_GW} ${BOND_DNS} ${BOND_DNS_SEARCH} ipv4.never-default no ipv4.ignore-auto-dns no nmcli con add autoconnect no type bond-slave con-name bond-slave-eth0 ifname eth0 master bond0 nmcli con add autoconnect no type bond-slave con-name bond-slave-eth1 ifname eth1 master bond0
#Enable bond nmcli con mod bond-slave-eth0 connection.autoconnect yes nmcli con up bond-slave-eth0
nmcli con mod bond-slave-eth1 connection.autoconnect yes nmcli con up bond-slave-eth1
nmcli con mod bond0 connection.autoconnect yes nmcli con up bond0
systemctl restart NetworkManager systemctl restart iptables
nmcli con | grep bond bond0 9942bdc6-df72-4723-b2ed-47a78e3a5c59 bond bond0 bond-slave-eth0 8b0fbbe1-a7f0-448c-8005-46d11599f57a 802-3-ethernet eth0 bond-slave-eth1 333dd1b9-15a4-4119-8e42-55ac3621a85d 802-3-ethernet eth1 *bond0 460dd9e8-bc0b-473e-9c89-41facda98b66 bond -- # Why this extra bond connections has been created?*
I'd appreciate your comments and suggestions to fix the issue.
Thanks,
Joe
To this day, on EL6, creating bonds always generates a spurious 'bond0' interface with no slaved interfaces. It was reported to red hat bugzilla ages ago but the issue was closed without resolution (sorry, I've been looking for the rhbz# but haven't found it yet).
digimer
On 07/07/16 05:36 PM, Digimer wrote:
On 07/07/16 05:21 PM, Joe Smithian wrote:
Hi All,
I see an unexpected beahviour from NetworkManager on CentOS 7.1. Using nmcli tool, I create a bond with two slaves as explained in the Red Hat 7.1 Networking guide. I enable slaves and master; bond works as expected. When I restart NetworkManager, it creates a new bond with the same name but not connected to any device. Two bonds with the same name is confusing for my other monitoring scripts. I'm wondering why a second bond is created? Is it a bug in NetworkManger?
#Create a bond with two slaves nmcli con add autoconnect no type bond con-name bond0 ifname bond0 nmcli con mod bond0 ipv6.method ignore ipv4.method manual ipv4.addresses ${BOND_IP}/${BOND_CIDR} ${BOND_GW} ${BOND_DNS} ${BOND_DNS_SEARCH} ipv4.never-default no ipv4.ignore-auto-dns no nmcli con add autoconnect no type bond-slave con-name bond-slave-eth0 ifname eth0 master bond0 nmcli con add autoconnect no type bond-slave con-name bond-slave-eth1 ifname eth1 master bond0
#Enable bond nmcli con mod bond-slave-eth0 connection.autoconnect yes nmcli con up bond-slave-eth0
nmcli con mod bond-slave-eth1 connection.autoconnect yes nmcli con up bond-slave-eth1
nmcli con mod bond0 connection.autoconnect yes nmcli con up bond0
systemctl restart NetworkManager systemctl restart iptables
nmcli con | grep bond bond0 9942bdc6-df72-4723-b2ed-47a78e3a5c59 bond bond0 bond-slave-eth0 8b0fbbe1-a7f0-448c-8005-46d11599f57a 802-3-ethernet eth0 bond-slave-eth1 333dd1b9-15a4-4119-8e42-55ac3621a85d 802-3-ethernet eth1 *bond0 460dd9e8-bc0b-473e-9c89-41facda98b66 bond -- # Why this extra bond connections has been created?*
I'd appreciate your comments and suggestions to fix the issue.
Thanks,
Joe
To this day, on EL6, creating bonds always generates a spurious 'bond0' interface with no slaved interfaces. It was reported to red hat bugzilla ages ago but the issue was closed without resolution (sorry, I've been looking for the rhbz# but haven't found it yet).
digimer
Found it:
https://bugzilla.redhat.com/show_bug.cgi?id=1245440
==== Neil Horman 2015-07-22 14:30:57 EDT
inserting the bonding module always creates the first bond interface, thats always how its been, and isn't a bug.
Status: NEW → CLOSED Resolution: --- → NOTABUG Last Closed: 2015-07-22 14:30:57 ====
digimer
Hi Neil,
Thanks for your comments. What's the purpose of creating spurious ‘bond0’? It's confusing. Is it anywhere documented? Every time I restart NetworkManager it creates another bond0!
Joe
On Thu, Jul 7, 2016 at 5:44 PM, Digimer lists@alteeve.ca wrote:
On 07/07/16 05:36 PM, Digimer wrote:
On 07/07/16 05:21 PM, Joe Smithian wrote:
Hi All,
I see an unexpected beahviour from NetworkManager on CentOS 7.1. Using nmcli tool, I create a bond with two slaves as explained in the
Red
Hat 7.1 Networking guide. I enable slaves and master; bond works as expected. When I restart NetworkManager, it creates a new bond with the same name
but
not connected to any device. Two bonds with the same name is confusing
for
my other monitoring scripts. I'm wondering why a second bond is created? Is it a bug in
NetworkManger?
#Create a bond with two slaves nmcli con add autoconnect no type bond con-name bond0 ifname bond0 nmcli con mod bond0 ipv6.method ignore ipv4.method manual
ipv4.addresses
${BOND_IP}/${BOND_CIDR} ${BOND_GW} ${BOND_DNS} ${BOND_DNS_SEARCH} ipv4.never-default no ipv4.ignore-auto-dns no nmcli con add autoconnect no type bond-slave con-name bond-slave-eth0 ifname eth0 master bond0 nmcli con add autoconnect no type bond-slave con-name bond-slave-eth1 ifname eth1 master bond0
#Enable bond nmcli con mod bond-slave-eth0 connection.autoconnect yes nmcli con up bond-slave-eth0
nmcli con mod bond-slave-eth1 connection.autoconnect yes nmcli con up bond-slave-eth1
nmcli con mod bond0 connection.autoconnect yes nmcli con up bond0
systemctl restart NetworkManager systemctl restart iptables
nmcli con | grep bond bond0 9942bdc6-df72-4723-b2ed-47a78e3a5c59 bond
bond0
bond-slave-eth0 8b0fbbe1-a7f0-448c-8005-46d11599f57a 802-3-ethernet
eth0
bond-slave-eth1 333dd1b9-15a4-4119-8e42-55ac3621a85d 802-3-ethernet
eth1
*bond0 460dd9e8-bc0b-473e-9c89-41facda98b66 bond -- # Why this extra bond connections has been created?*
I'd appreciate your comments and suggestions to fix the issue.
Thanks,
Joe
To this day, on EL6, creating bonds always generates a spurious 'bond0' interface with no slaved interfaces. It was reported to red hat bugzilla ages ago but the issue was closed without resolution (sorry, I've been looking for the rhbz# but haven't found it yet).
digimer
Found it:
https://bugzilla.redhat.com/show_bug.cgi?id=1245440
==== Neil Horman 2015-07-22 14:30:57 EDT
inserting the bonding module always creates the first bond interface, thats always how its been, and isn't a bug.
Status: NEW → CLOSED Resolution: --- → NOTABUG Last Closed: 2015-07-22 14:30:57 ====
digimer
-- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education? _______________________________________________ CentOS mailing list CentOS@centos.org https://lists.centos.org/mailman/listinfo/centos
On Mon, 11 Jul 2016 16:47, Joe Smithian wrote:
Hi Neil,
Thanks for your comments. What's the purpose of creating spurious ‘bond0’? It's confusing. Is it anywhere documented? Every time I restart NetworkManager it creates another bond0!
Joe
[snip]
For me the soution was to create a script that removes the "bond0" when NetworkManager is stopped, I have it integrated into the NetworkManager.service via a drop-in scriptlet into the dir "/etc/systemd/system/NetworkManager.service.d/"
here is my scriptlet: /etc/systemd/system/NetworkManager.service.d/remove-bond0-at-stop
[code] # remove "bond0" interface at stop, it will be re-created at start [Service] ExecStopPost=/usr/bin/bash /path/to/shellscript [/code]
sadly the shell script has been lost since, with the move to a machine with 10GbE interfaces.
- Yamaban.