Hello! I hope it is okay to ask here, as this is primarily a hardware question, and not really an oWRT issue.
I am using a BPI-R4 as my main router, running the latest release candidate at the time of writing this (OpenWrt 24.10.0-rc4 r28211-d55754ce0d). I had been using older versions before, with the same issue. Originally I was supplying the BPI-R4 with the weak factory issued 24W PSU, I have switched to a 65W USB-C PD PSU for the time being to exclude one possible cause.
My desktop PC ("Killer" E3100G 2.5G NIC) is currently connected to one of the SFP+ ports with a 10Gtek ASF-10G2-T 10Gbit Ethernet transceiver (Marvell AQR113C, according to the manufacturer). It works, as in I can get stable 2.35 Gbit with iperf, but every now and then (sometimes works for hours, sometimes crashes every 10 seconds) the link disconnects and reconnects.
Sys log of such an event:
Sat Dec 28 16:40:00 2024 daemon.notice netifd: Network device 'eth1' link is down
Sat Dec 28 16:40:00 2024 kern.info kernel: [ 3777.108737] mtk_soc_eth 15100000.ethernet eth1: Link is Down
Sat Dec 28 16:40:00 2024 kern.info kernel: [ 3777.114719] br-lan: port 1(eth1) entered disabled state
Sat Dec 28 16:40:06 2024 kern.info kernel: [ 3782.935451] mtk_soc_eth 15100000.ethernet eth1: Link is Up - 2.5Gbps/Full - flow control rx/tx
Sat Dec 28 16:40:06 2024 kern.info kernel: [ 3782.935477] br-lan: port 1(eth1) entered blocking state
Sat Dec 28 16:40:06 2024 kern.info kernel: [ 3782.949295] br-lan: port 1(eth1) entered forwarding state
Sat Dec 28 16:40:06 2024 daemon.notice netifd: Network device 'eth1' link is up
ip -s -s link show output shows CRC RX errors (not counting up at every disconnect) and transns TX errors (counting up at every disconnect):
root@nexus:~# ip -s -s link show eth1
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq master br-lan state UP mode DEFAULT group default qlen 1000
link/ether ce:57:c4:7f:88:de brd ff:ff:ff:ff:ff:ff
RX: bytes packets errors dropped missed mcast
11890970 66116 0 151 0 0
RX errors: length crc frame fifo overrun
0 4 0 0 0
TX: bytes packets errors dropped carrier collsns
66402906 76715 0 0 0 0
TX errors: aborted fifo window heartbt transns
0 0 0 0 119
I am using a 1m CAT7 cable to connect the transceiver to the patch panel. The patch panel is a CAT6a model. There is an approximate 10m of CAT7 cable inside the walls connected on the other end to a CAT6a outlet. From the outlet to my PX I am using another meter of CAT7 cable.
Is there any way to further debug this issue? Is it a transceiver issue? A cable issue? A hardware issue of the BPI-R4? I can't physically move my desktop over to the other room, and I can't move the router to my room without shutting my whole network down for good, so I can't really test with a short direct connection.
The metrics reported by the transceiver seem uneventful:
