Networking doesn't resume if SFP cable is plugged out and plugged in back.

[grvjn]

Hi, 

We were testing the SFP port, and we plugged out the sfp cable, just the cable not whole module and plugged it back in. We observed -

1. Networking resumes like ssh start working and other interface resumes if we plug back the sfp cable before 3 minutes.

2. Beyond 3 minutes networking doesn't resume. 

Is it the expected behavior, how can we restart networking beyond 3 minutes. 

Thanks and Regards,

Gaurav Jain 

[stadium_aquino]

The SFP port isn't described to Linux. It doesn't know that the module has been (un)plugged. 6.3 should support it when the dpaa phylink conversion lands, but you will need to modify your dts to add the sfp module.

[grvjn]

Hi,

It is a custom board, processor is ls1046a - arm8. SFP port is 10 gbps. 

Regards,

Gaurav

[yipingwang]

When the SFP cable is plugged out, the signal will be lost, and the networking traffic should be stopped.
The ping commands will show icmp packets loss if used "ping <ip_address>".

For ssh, if the board as client, you can check your server configuration "ClientAliveInterval 180" in "/etc/ssh/sshd_config".
it means ssh server will send a packet each 180s to keep the connective.
If the networking resumes within 180s, the ssh connection will be kept.

[grvjn]

When the cable is re-plugged, what is expected -

a. Networking will resume, or

b. Reboot is required to bring up networking.

[yipingwang]

1. Networking will resume, or

// for SSH connection, it can be resumed in 180s (depended local configurations)

1. Reboot is required to bring up networking.

// What Linux distribution did customer use? if it is ubuntu, customer can use netplan, here is a example.

I assume there are two ports in linux.

root@localhost:~# ls-listni

dprc.1/dpni.1 (interface: eth1, end point: dpmac.2)

dprc.1/dpni.0 (interface: eth2, end point: dpmac.17)

1. create a yaml file as below.

root@localhost:~# cat /etc/netplan/01-netplan_dpaa2_cfg.yaml

network:

    version: 2

    renderer: networkd

    ethernets:

        eth1:

            addresses: [10.0.0.4/24]

            gateway4: 10.0.0.1

        eth2:

            addresses: [10.0.2.5/24]

            gateway4: 10.0.2.1

And execute "netplan --debug apply". you can see eth1 and eth2 had been link up.

root@localhost:~# ifconfig

eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500

        inet 10.0.0.4  netmask 255.255.255.0  broadcast 10.0.0.255

        inet6 fe80::204:9fff:fe06:fbd7  prefixlen 64  scopeid 0x20<link>

        ether 00:04:9f:06:fb:d7  txqueuelen 1000  (Ethernet)

        RX packets 0  bytes 0 (0.0 B)

        RX errors 0  dropped 0  overruns 0  frame 0

        TX packets 11  bytes 866 (866.0 B)

        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

eth2: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500

        inet 10.0.2.5  netmask 255.255.255.0  broadcast 10.0.2.255

        inet6 fe80::204:9fff:fe06:fbdc  prefixlen 64  scopeid 0x20<link>

        ether 00:04:9f:06:fb:dc  txqueuelen 1000  (Ethernet)

        RX packets 90  bytes 11760 (11.7 KB)

        RX errors 0  dropped 0  overruns 0  frame 0

        TX packets 11  bytes 866 (866.0 B)

        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536

        inet 127.0.0.1  netmask 255.0.0.0

        inet6 ::1  prefixlen 128  scopeid 0x10<host>

        loop  txqueuelen 1000  (Local Loopback)

        RX packets 640  bytes 45680 (45.6 KB)

        RX errors 0  dropped 0  overruns 0  frame 0

        TX packets 640  bytes 45680 (45.6 KB)

        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

root@localhost:~#

After reboot, eth1 and eth2 will be link up by netplan.

[yipingwang]

Which processor are you using?

Do you use NXP demo board or your custom board?