XMG1915-10E & TP-link TX401

epierre

Hi All,

I have recurrent issues where the switch disconnects the network card, and it happens both under windows 11 and Linux.

Bandwith control, loop, storm, port isolation and flow are disabled for this port,

RTSP, LLDP, Link are enabled, 10G forced (auto before still with issue).

Logs are quite scarce, here is what I have under windows when it finally connects:

2024-10-11 11:02:23 Notice SW2 ddmiInterfaceAbnormal RX power state at port 9 - Port9 has been cleared

2024-10-11 11:01:12 Debug SW2 interfaceInterfacePort 9 - Port9 link up 10G/F

2024-10-11 10:47:23 Notice SW2 ddmiInterfaceRxPower Under Alarm Low Threshold(-17.01),Port 9 - Port9, CurValue:negative infinity(0mW)

2024-10-11 10:45:09 Debug SW2 interfaceInterfacePort 9 - Port9 link down

epierre

More info: suddently the link goes up without doing anything on my side:

Oct 11 16:02:37 kernel: atlantic 0000:07:00.0 enp7s0: atlantic: link change old 0 new 10000

2024-10-11 16:02:39 Debug SW2 interfaceInterface Port 9 - Port9 link up 10G/F

When OK DDMI says:

Voltage(V) 3.18
TX Power(dbm) -3.01
TX Bias(mA) 6.00
Temperature(C) 62.01
RX Power(dbm) -3.98
if not, RX is 0

Some more infos from linux logs when going down:

Oct 11 17:54:40 kernel: IPv4: martian source 192.168.xx from 192.168.xx, on dev enp7s0
Oct 11 17:54:40 kernel: ll header: 00000000: ff ff ff ff ff ff e4 60 17 47 91 0e 08 06
Oct 11 17:54:41 kernel: IPv4: martian source 192.168.xy from 192.168xy on dev wlo1
Oct 11 17:54:41 kernel: ll header: 00000000: ff ff ff ff ff ff 5c e9 31 07 e2 3f 08 06
Oct 11 17:54:46 kernel: atlantic 0000:07:00.0 enp7s0: atlantic: link change old 10000 new 0

then up again:

Oct 11 17:55:05 kernel: atlantic 0000:07:00.0 enp7s0: atlantic: link change old 0 new 10000

and down again:

Oct 11 17:56:06 kernel: IPv4: martian source 192.168.86.133 from 192.168.xx, on dev enp7s0
Oct 11 17:56:06 kernel: ll header: 00000000: ff ff ff ff ff ff e4 60 17 47 91 0e 08 06
Oct 11 17:56:53 kernel: atlantic 0000:07:00.0 enp7s0: atlantic: link change old 10000 new 0

Zyxel_Melen

Hi @epierre,

What SFP module did you use?

epierre

It IS a 10gtek asf-10g2-t

chipset Marvell AQR113C

epierre

Other things I see: windows or linux after a suspend, the ethernet link never goes up again

Zyxel_Melen

Hi @epierre,

From the logs you provided, it is more likely due to the SFP module. ZYXEL switch products are recommended to be used exclusively with ZYXEL brand modules to ensure optimal performance and reliability.Using modules from third-party sources can potentially lead to decreased network performance and reliability, such as packet loss and connectivity issues, which also cannot be guaranteed by ZYXEL. ZYXEL may not offer technical support for networking issues that are identified as being caused by IOP with third-party modules.

Below are the SFP modules we recommended:

Zyxel_Melen

Hi @epierre,

In addition, the cable length limit of the RJ-45 10G SPF+ module is 30 meters. Please make sure the length of your cable is shorter than 30 meters.

epierre

Hello,

BTW cable is less than 10m.

Searched the module above, not sold on zyxel website … surely too confidential...

This is m'y first zyxel, not a great experience (not to speak about nebula forcing everytime to buy useless modules for this equipment segment and proprietary …), I'm regreting my purchase and the functionalities I had on NETGEAR and alike.

epierre

well going from firmware stable to latest seems to fix the issue… what a pain !