I've had a few reboots where networking wouldn't start right away, but that's just until the bond reinitializes the slaves. Surely, a network restart helps, but it would come up anyway. You might have something like that in dmesg in such a boot :
Code:
[ 7.253322] bonding: bond0: doing slave updates when interface is down.
Then it detects the first interface, but it seems down ( the switch didn't up the port fast enough i guess ) :
Code:
[ 7.253329] bonding: bond0: enslaving eth0 as a backup interface with a down link
Then it detected the second link :
Code:
[ 7.253354] bonding: bond0: Adding slave eth1.
Then it took a few more probes to sort it out :
Code:
[ 7.253356] bonding bond0: master_dev is not up in bond_enslave
[ 7.310082] alloc irq_desc for 62 on node -1
[ 7.310086] alloc kstat_irqs on node -1
[ 7.310100] bnx2 0000:05:00.0: irq 62 for MSI/MSI-X
[ 7.440012] bnx2: eth1: using MSI
[ 7.442014] bonding: bond0: enslaving eth1 as a backup interface with a down link.
[ 7.442010] bonding: bond0: Setting eth0 as primary slave.
[ 7.603196] bonding: bond0: Setting eth0 as primary slave.
[ 7.606110] device bond0 entered promiscuous mode
[ 7.609903] ADDRCONF(NETDEV_UP): bond0: link is not ready
[ 8.889748] bnx2: eth0 NIC Copper Link is Up, 100 Mbps full duplex
[ 8.901886] bonding: bond0: link status definitely up for interface eth0.
[ 8.901891] bonding: bond0: making interface eth0 the new active one.
[ 8.901895] device eth0 entered promiscuous mode
[ 8.901919] bonding: bond0: first active interface up!
[ 8.903921] ADDRCONF(NETDEV_CHANGE): bond0: link becomes ready
[ 8.903987] br0: port 1(bond0) entering forwarding state
[ 9.168875] bnx2: eth1 NIC Copper Link is Up, 100 Mbps full duplex
[ 9.201885] bonding: bond0: link status definitely up for interface eth1.
While in previous ubuntu's i didn't had such "challenges" with bonding, it's still working properly in lucid...just a tad bit weirder.
Bookmarks