CentOS - Sep 2019 - 7.7.1908, interface bonding, and default route

Hi!

I just upgraded a machine to 7.7.1908 and the default route is not being
set on boot. This particular server has a bonded interface, and the
corresponding configuration for the master is (
/etc/sysconfig/network-scripts/ifcfg-bond0):

  TYPE=Bond
  BOOTPROTO=none
  DEFROUTE=yes
  IPV4_FAILURE_FATAL=yes
  NAME=bond0
  DEVICE=bond0
  ONBOOT=yes
  IPADDR=10.3.20.131
  PREFIX=24
  GATEWAY=10.3.20.1
  DNS1=10.3.2.8
  BONDING_MASTER=yes
  BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer2 miimon=100"

The slaves (two of them) are configured like

  TYPE=Ethernet
  BOOTPROTO=none
  NAME=bond0-slave0
  DEVICE=em3
  ONBOOT=yes
  MASTER=bond0
  SLAVE=yes

After booting, the routing table is

  10.3.20.0/24 dev bond0 proto kernel scope link src 10.3.20.131 metric 300

with no default route configured (manually adding it will work, of course.)

This machine worked perfectly before, and it did during previous upgrades
and reboots. Has anyone ever had this problem before?

Thanks,
Carlos.

On 20/09/2019 04:55, Carlos A. Carnero Delgado wrote:
> Hi!
>
> I just upgraded a machine to 7.7.1908 and the default route is not being
> set on boot. This particular server has a bonded interface, and the
> corresponding configuration for the master is (
> /etc/sysconfig/network-scripts/ifcfg-bond0):
>
>    TYPE=Bond
>    BOOTPROTO=none
>    DEFROUTE=yes
>    IPV4_FAILURE_FATAL=yes
>    NAME=bond0
>    DEVICE=bond0
>    ONBOOT=yes
>    IPADDR=10.3.20.131
>    PREFIX=24
>    GATEWAY=10.3.20.1
>    DNS1=10.3.2.8
>    BONDING_MASTER=yes
>    BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer2 miimon=100"
>
> The slaves (two of them) are configured like
>
>    TYPE=Ethernet
>    BOOTPROTO=none
>    NAME=bond0-slave0
>    DEVICE=em3
>    ONBOOT=yes
>    MASTER=bond0
>    SLAVE=yes
>
> After booting, the routing table is
>
>    10.3.20.0/24 dev bond0 proto kernel scope link src 10.3.20.131 metric
300
>
> with no default route configured (manually adding it will work, of course.)
>
> This machine worked perfectly before, and it did during previous upgrades
> and reboots. Has anyone ever had this problem before?
>
I have a similar set up to you, and just did the upgrade to 1908, I 
didn't experience the problem you had, I can't see anything out of the 
ordinary in your network files.

I originally set up my bond using nmtui

Here is my configuration:

#Bond:

BONDING_OPTS="ad_select=stable all_slaves_active=0 arp_all_targets=any 
downdelay=0 fail_over_mac=none lp_interval=1 miimon=100 min_links=0 
mode=802.3ad num_grat_arp=1 num_unsol_na=1 primary_reselect=always 
resend_igmp=1 updelay=0 use_carrier=1 xmit_hash_policy=layer2"
TYPE=Bond
BONDING_MASTER=yes
PROXY_METHOD=none
BROWSER_ONLY=no
BOOTPROTO=none
IPADDR=a.b.c.249
PREFIX=24
GATEWAY=a.b.c.1
DNS1=a.b.d.253
DNS2=a.b.d.4
DNS3=a.b.d.12
DOMAIN=redacted.net
DEFROUTE=yes
IPV4_FAILURE_FATAL=no
IPV6INIT=yes
NAME="Bond connection 1"
UUID=4ccacb42-dfad-484f-8168-b78c70a66c8d
DEVICE=nm-bond
ONBOOT=yes
IPV6_AUTOCONF=yes
IPV6_DEFROUTE=yes
IPV6_FAILURE_FATAL=no
IPV6_ADDR_GEN_MODE=stable-privacy

#Channel Interface example

TYPE=Ethernet
NAME="Ethernet connection 1"
UUID=5a961074-d502-49a3-ae3b-b6850a990e86
DEVICE=enp7s0f0
ONBOOT=yes
MASTER=nm-bond
SLAVE=yes
MASTER_UUID=4ccacb42-dfad-484f-8168-b78c70a66c8d
> Thanks,
> Carlos.
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> https://lists.centos.org/mailman/listinfo/centos
-- 
Giles Coochey

El vie., 20 de sep. de 2019 a la(s) 06:16, Giles Coochey (giles at coochey.net)
escribi?:
> I have a similar set up to you, and just did the upgrade to 1908, I
> didn't experience the problem you had, I can't see anything out of
the
> ordinary in your network files.
>
I have reviewed the configuration several times now, and still can't see if
there's anything wrong with it. What I found is that, after the system has
booted up, if I systemctl restart network the default route does gets
applied.

Thanks a lot for your input!