NetworkManager timeout and "ip-config-unavailable" on Ethernet

The probable cause was a defective Ethernet port on the router. I didn't find the evidence to support this until 30 April 2015, and I didn't make the connection until 10 August 2015.

What I Should Have Done

A few additional troubleshooting steps:

• I should have run sudo ethtool -S eth0 to see if there were any errors in the network interface controller (NIC) statistics.

  Since I was able to send but not receive over Ethernet, I might have seen something like the rx_crc_errors statistic increment on each receive attempt.

• I should have tried plugging in the router end of the Ethernet cable into a different router port.

  Sure, I plugged in the computer end of the Ethernet cable into two different computers, but that led me to believe that a computer (Craptop) was faulty because the other (Deltique) was able to connect to the router.

• On the working computer, Deltique, I should have run sudo ethtool eth0 to see if a good link was established.

  Deltique has a 1Gbps capacity, and so does the router, so ethtool should have shown Speed: 1000Mb/s and Duplex: Full. I would later discover on 30 April 2015 that the defective port would have caused ethtool to return different, more concerning values.

• I should have run a speed benchmark to test the Ethernet link to make sure that I was getting the full performance expected between Deltique and the router.

  Had I done a speed test, I likely would have seen a 1MiB/s transfer speed, indicating a link problem, even on Deltique.

• To rule out a cable problem, I should have tried using a different Gigabit Ethernet cable.

  It turns out that the cable was fine, but this is a good troubleshooting step anyway.

Evidence and Explanation

On 30 April 2015, I noticed that my trusty rsync -avzHXShPs command was maxing out at about 1MiB/s. I ruled out other traffic by monitoring iftop and by attempting an Internet download on a different computer, which transferred data ten times faster.

I checked the network link with ethtool:

root@node51 [~]# ethtool eth0
Settings for eth0:
        Supported ports: [ TP MII ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Half 1000baseT/Full
        Supported pause frame use: No
        Supports auto-negotiation: Yes
        Advertised link modes:  10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Half 1000baseT/Full
        Advertised pause frame use: Symmetric Receive-only
        Advertised auto-negotiation: Yes
        Link partner advertised link modes:  10baseT/Half
        Link partner advertised pause frame use: No
        Link partner advertised auto-negotiation: No
        Speed: 10Mb/s
        Duplex: Half
        Port: MII
        PHYAD: 0
        Transceiver: internal
        Auto-negotiation: on
        Supports Wake-on: pumbg
        Wake-on: g
        Current message level: 0x00000033 (51)
                               drv probe ifdown ifup
        Link detected: yes

Aha! 10Mbps at half duplex. That's not normal and is very slow. For reference, this was the expected (good) output, captured by switching Ethernet ports on the router:

root@node51 [~]# ethtool eth0
Settings for eth0:
        Supported ports: [ TP MII ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Half 1000baseT/Full
        Supported pause frame use: No
        Supports auto-negotiation: Yes
        Advertised link modes:  10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Half 1000baseT/Full
        Advertised pause frame use: Symmetric Receive-only
        Advertised auto-negotiation: Yes
        Link partner advertised link modes:  10baseT/Half 10baseT/Full
                                             100baseT/Half 100baseT/Full
                                             1000baseT/Full
        Link partner advertised pause frame use: Symmetric
        Link partner advertised auto-negotiation: Yes
        Speed: 1000Mb/s
        Duplex: Full
        Port: MII
        PHYAD: 0
        Transceiver: internal
        Auto-negotiation: on
        Supports Wake-on: pumbg
        Wake-on: g
        Current message level: 0x00000033 (51)
                               drv probe ifdown ifup
        Link detected: yes

The computer and the router tried to negotiate slower and slower speeds until a reliable one could be established. The slowest one, 10BASE-T half duplex, was picked.

Forcing the full 1000BASE-T full duplex (ethtool -s eth0 speed 1000 duplex full) expected performance caused the affected computer to drop from the network, strongly suggesting that the router didn't want to talk on a faster link.

Craptop, which is limited to 100Mbps full duplex in the first place, probably didn't have a chance to negotiate 10Mbps half duplex.

Now that I have identified the bad Ethernet port, all I need to do is not use that port or get a new router.

On 27 July 2015, I returned the defective router after installing a new router.

Today, Craptop is running on a fault-tolerant network bond with a 100Mbps full duplex Ethernet connection:

root@node52 [~]# cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)

Bonding Mode: fault-tolerance (active-backup)
Primary Slave: eth0 (primary_reselect always)
Currently Active Slave: eth0
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: wlan0
MII Status: up
Speed: Unknown
Duplex: Unknown
Link Failure Count: 3
Permanent HW addr: 00:1f:3a:55:dc:0d
Slave queue ID: 0

Slave Interface: eth0
MII Status: up
Speed: 100 Mbps
Duplex: full
Link Failure Count: 3
Permanent HW addr: 00:1e:68:32:4a:b5
Slave queue ID: 0