1
I'm trying to troubleshoot boot times for a relatively modern machine with a SSD. The machine runs Ubuntu Server 14.04.3 LTS and it has two ethernet ports. Network Manager is not installed. All device are Gigabit ethernet capable, and Cat6 cabling is used throughout. The cable lengths from the server to the switch are 6 or 10 feet.
When only eth0
was in use with DHCP, the machine would boot in about 2.5 to 3 seconds.
I enabled eth1
and added a br0
to support host-bridging for virtualization. The VMs use the second ethernet interface, and none of the VMs auto-start. Now the boot times are 30 seconds or so.
How do I troubleshoot this?
Or how do I fix it (if it can be determined from the limited information I have)?
/etc/networks/interfaces
Changing the spanning tree protocol state (on or off) makes no difference. Commenting out the bridge drops the boot time to about 5 seconds.
$ cat /etc/network/interfaces
...
# Primary network interface
auto eth0
iface eth0 inet dhcp
# Secondary network interface
iface eth1 inet manual
pre-up ifconfig $IFACE up
post-down ifconfig $IFACE down
# Bridged interface for virtualization
auto br0
iface br0 inet dhcp
bridge_ports eth1
bridge_stp on
bridge_fd 0
bridge_maxwait 0
dmesg output
$ dmesg | egrep -i "(eth|r1869)"
[ 1.580017] r8169 Gigabit Ethernet driver 2.3LK-NAPI loaded
[ 1.581643] r8169 0000:02:00.0 eth0: RTL8168evl/8111evl at 0xffffc900018fc000, 00:e0:4c:68:03:5e, XID 0c900800 IRQ 46
[ 1.582988] r8169 0000:02:00.0 eth0: jumbo features [frames: 9200 bytes, tx checksumming: ko]
[ 1.583755] r8169 Gigabit Ethernet driver 2.3LK-NAPI loaded
[ 1.585640] r8169 0000:03:00.0 eth1: RTL8168evl/8111evl at 0xffffc90005b98000, 00:e0:4c:68:03:5f, XID 0c900800 IRQ 48
[ 1.587100] r8169 0000:03:00.0 eth1: jumbo features [frames: 9200 bytes, tx checksumming: ko]
[ 3.086938] r8169 0000:02:00.0 eth0: link down
[ 3.086956] r8169 0000:02:00.0 eth0: link down
[ 5.463989] r8169 0000:02:00.0 eth0: link up
[ 7.138820] device eth1 entered promiscuous mode
[ 7.248670] r8169 0000:03:00.0 eth1: link down
[ 7.248687] r8169 0000:03:00.0 eth1: link down
[ 25.169258] r8169 0000:03:00.0 eth1: link up
[ 25.169742] br0: port 1(eth1) entered listening state
[ 25.169748] br0: port 1(eth1) entered listening state
[ 27.171555] br0: port 1(eth1) entered learning state
[ 29.174455] br0: port 1(eth1) entered forwarding state
ethtool
$ sudo ethtool eth0 | egrep -i "(speed|duplex|neg)"
Speed: 1000Mb/s
Duplex: Full
Auto-negotiation: on
$ sudo ethtool eth1 | egrep -i "(speed|duplex|neg)"
Speed: 100Mb/s
Duplex: Full
Auto-negotiation: on
*[ 5.463989] r8169 0000:02:00.0 eth0: link up* Probably it is either other end of the cable or cable itself – aaaaa says reinstate Monica – 2016-02-04T08:48:57.153
1Thanks again @grawity. You we right about the cable: with two new Cat6 cables in place, it takes about 11 seconds to boot (down from 25 to 30). 6 seconds of the new boot time appears to be due to the bridge. I'll look at tuning the bridge in another question. – jww – 2016-02-06T20:29:05.737