Bug 343844

Summary: vlan on bonding fails to start cleanly
Product: [openSUSE] openSUSE 10.3 Reporter: Mike Atkin <mike>
Component: NetworkAssignee: Marius Tomaschewski <mt>
Status: RESOLVED FIXED QA Contact: E-mail List <qa-bugs>
Severity: Normal    
Priority: P5 - None CC: chrubis, jdelvare, reitenbach
Version: Final   
Target Milestone: ---   
Hardware: Other   
OS: openSUSE 10.3   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: proposed patch

Description Mike Atkin 2007-11-23 16:45:37 UTC
I'm trying to get the following networking setup running:

ifcfg-eth0:

STARTMODE=off

ifcfg-eth1:

STARTMODE=off

ifcfg-bond0:

STARTMODE='auto'
BOOTPROTO=static
BONDING_MASTER=yes
BONDING_MODULE_OPTS='mode=4 miimon=100'
BONDING_SLAVE_0=eth0
BONDING_SLAVE_1=eth1

ifcfg-vlan3:

STARTMODE=auto
IPADDR=192.168.11.3
NETMASK=255.255.255.0
ETHERDEVICE=bond0

+ a few more similar vlans


When the interfaces are brought up via the init script, the script hangs waiting for mandatory device 'bond0' to appear.  After timing out, the script continues and starts the bonding interface and the vlan interface fine, but exits with an error due to the earlier timeout.  It appears that the init script expects the ETHERDEVICE variable to be a physical interface.

This first issue can be worked around by manually setting MANDATORY_DEVICES="eth0 eth1" in /etc/sysconfig/network/config thereby bypassing the automatic detection.

The scripts also appear to be confused by the /sys/class/net/bonding_masters file on a restart, thinking this is also a networking interface, and then go on to probe the vlan devices for their bonding slaves!?!:

# rcnetwork restart
Shutting down network interfaces:
    bond0     is still used from interfaces vlan2
vlan3
vlan4
vlan5
    vlan2
cat: /sys/class/net/vlan2/bonding/slaves: No such file or directory
    vlan3
cat: /sys/class/net/vlan3/bonding/slaves: No such file or directory
    vlan4
cat: /sys/class/net/vlan4/bonding/slaves: No such file or directory
    vlan5
cat: /sys/class/net/vlan5/bonding/slaves: No such file or directory
    bond0     now going down itself
    bond0
    bond0     removed slave : eth0
    bond0     removed slave : eth1                                    done
    vlan2                                                             done
    vlan3                                                             done
    vlan4                                                             done
    vlan5                                                             done
    bonding_masters
Interface not available and no configuration found.                   done
    eth0      device: Broadcom Corporation NetXtreme BCM5714 Gigabit Edonenet (rev a3)
    eth1      device: Broadcom Corporation NetXtreme BCM5714 Gigabit Edonenet (rev a3)
Shutting down service network  .  .  .  .  .  .  .  .  .  .  .  .  .  done
Setting up network interfaces:
    lo
    lo        IP address: 127.0.0.1/8
Checking for network time protocol daemon (NTPD):                     doneed
Interface bonding_masters is not available                            failed
    eth0      device: Broadcom Corporation NetXtreme BCM5714 Gigabit Ethernet (rev a3)
    eth0      Startmode is 'off'                                      skipped
    eth1      device: Broadcom Corporation NetXtreme BCM5714 Gigabit Ethernet (rev a3)
    eth1      Startmode is 'off'                                      skipped
    bond0
    bond0     enslaved interface: eth0
    bond0     enslaved interface: eth1
Checking for network time protocol daemon (NTPD):                     doneed
    vlan2
Checking for network time protocol daemon (NTPD):                     doneed
    vlan3
    vlan3     IP address: 192.168.11.3/24
Checking for network time protocol daemon (NTPD):                     doneed
    vlan4
    vlan4     IP address: 192.168.20.17/24
Checking for network time protocol daemon (NTPD):                     doneed
    vlan5
    vlan5     IP address: 192.168.23.3/24
Checking for network time protocol daemon (NTPD):                     doneed
Setting up service network  .  .  .  .  .  .  .  .  .  .  .  .  .  .  done




Something here also confuses the network configuration yast module.  When it tries to restart the networking after a configuration change it sometimes leaves all interfaces down and removes the default route from the config file.
Comment 1 Cyril Hrubis 2007-11-26 15:24:22 UTC
Please open new bug for yast and keep policy one problem per report.
Comment 2 Marius Tomaschewski 2007-12-18 13:51:43 UTC
Created attachment 188051 [details]
proposed patch

To apply with:
  patch -p1 < rcnetwork-bridge-bonding.dif /etc/init.d/network
Comment 3 Marius Tomaschewski 2007-12-18 13:54:24 UTC
Mike, I think this is just because of the start order in /etc/init.d/network.

Can you try out if it works when you apply the patch from comment 2? Thanks!
Comment 4 Mike Atkin 2007-12-18 18:13:33 UTC
I won't have any equipment available until mid Jan to test this now, but I'll let you know what happens as soon as I do.
Comment 5 Marius Tomaschewski 2008-04-25 10:07:05 UTC
fixed in last update (09-Apr-2008)
Comment 6 Marius Tomaschewski 2008-04-25 10:11:22 UTC
*** Bug 343600 has been marked as a duplicate of this bug. ***