With bond 802.3ad one interfaces sometimes fall back to 10 Mpbs

rabudde · Jul 24th 2020

I'm using OMV 4.1.35 along with an ASRockRack C236 WSI. This motherboard has 2 NICs and is attached to Unifi network switch. Normally, both links where up with 1000 Mpbs.

Code

# omv-confdbadm read --prettify conf.system.network.interface
[
    {
        "address": "192.168.1.20",
        "address6": "",
        "bonddowndelay": 200,
        "bondmiimon": 100,
        "bondmode": 4,
        "bondprimary": "",
        "bondupdelay": 200,
        "comment": "",
        "devicename": "bond0",
        "dnsnameservers": "192.168.1.1",
        "dnssearch": "geheim",
        "gateway": "192.168.1.1",
        "gateway6": "",
        "method": "static",
        "method6": "manual",
        "mtu": 0,
        "netmask": "255.255.255.0",
        "netmask6": 64,
        "options": "",
        "slaves": "enp0s31f6,enp3s0",
        "type": "bond",
        "uuid": "a60a3150-33a3-4e83-a070-b4505725ac69",
        "vlanid": 1,
        "vlanrawdevice": "",
        "wol": false,
        "wpapsk": "",
        "wpassid": ""
    }
]

Display More

I cannot find a reason, but sometimes the interface enp0s31f6 falls back to 10 Mpbs which gives trouble for some clients accessing NAS. This happened rarely but only this interface is affected, the other NIC is fine alltime.

Code

[878275.310646] e1000e: enp0s31f6 NIC Link is Down
[878275.310660] igb 0000:03:00.0 enp3s0: igb: enp3s0 NIC Link is Down
[878275.310900] e1000e 0000:00:1f.6 enp0s31f6: speed changed to 0 for port enp0s31f6
[878275.311000] igb 0000:03:00.0 enp3s0: speed changed to 0 for port enp3s0
[878275.334757] bond0: link status definitely down for interface enp0s31f6, disabling it
[878275.334814] bond0: link status definitely down for interface enp3s0, disabling it
[878275.334818] bond0: now running without any active interface!
[878310.034521] igb 0000:03:00.0 enp3s0: igb: enp3s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
[878310.098013] bond0: link status up for interface enp3s0, enabling it in 0 ms
[878310.098220] bond0: link status definitely up for interface enp3s0, 1000 Mbps full duplex
[878310.098233] bond0: first active interface up!
[878312.170422] e1000e: enp0s31f6 NIC Link is Up 10 Mbps Full Duplex, Flow Control: None
[878312.170424] e1000e 0000:00:1f.6 enp0s31f6: 10/100 speed: disabling TSO
[878312.178050] bond0: link status up for interface enp0s31f6, enabling it in 200 ms
[878312.178052] bond0: invalid new link 3 on slave enp0s31f6
[878312.386248] bond0: link status definitely up for interface enp0s31f6, 10 Mbps full duplex

Display More

It seems when link to switch is broken (i.e. when switch has a power failure - the NAS itself has an UPS), the autonegotiation fails when link comes up again. But I'm not sure if this is the real reason.

When this happens, I have to SSH into the NAS and run

Code

ifdown bond0 && ifup bond0

After that everything is fine again. dmesg gives me the following.

I cannot believe this should be a fault of OMV, but I don't know where I have to begin my research. Or has anyone an idea what I can do to automatically recover the interface to full speed?

Participate now!