WAN Interface Losing Link

stephenw10

Yup swap the ports, see if the fault follows the port.

Try putting a switch in between the WAN and modem, see if one side still looses link and which side it is.

Steve

Scallica

@stephenw10 Thanks, I will try the switch test. When the issue was happening, I removed the Ethernet cable from the WAN port, waited a few seconds, then plugged the cable back in. The issue persisted.

Scallica

The issue happened again today. The WAN link light was flashing on/off. I removed the WAN cable and plugged it into a gigabit switch. The link light remained on. I reconnected the WAN cable and then changed the WAN interface setting from autoselect to 1000base-T full duplex. The link light remained on.

bingo600

Watch out if you are "Only forcing one end of a link" , some (if not all) chipsets.
Will fallback to HDPX if the other end will not negotiate (is forced)

I always recommend to "force to the same in both ends" if possible.

/Bingo

Gertjan

@Scallica said in WAN Interface Losing Link:

Gateway alarm: WAN_DHCP .... Loss:28%)

and the threshold dpinger (the gateway reacability tester) is "pulling" the plug when ? 30 %

Your WAN goes down probably because the "Gateway alarm" (dpinger) is instructed to do so : to many ICMP packets are lost, the connection is considered bad. It resets your WAN.

Try checking this option :

Add to you do-check-list :
Why are there so many ICMP packets get lost ?
What happens if you instruct dpinger (gateway alarm tester) NOT to pull the plug : doing nothing when many packets are lost
Like this :

Scallica

I always recommend to "force to the same in both ends" if possible.

I agree, but in this case, the other end is a Verizon FiOS ONT. There have been no further issues in the last 30 days.

Scallica

@bingo600 As a test, I switched the Speed and Duplex setting back to autonegotiate. Immediately, the WAN interface started going up and down. The link lights would stay on for two seconds and off for one second. I think it is safe to say there is an negotiation issue between the WAN interface on the pfSense box and the Ethernet interface of the Verizon ONT.

bingo600

@Scallica

Hmm my bad (i need glasses)
I missed the 1000Mb .. Afaik 1000-TX can only run Fdpx

So setting 1000-Fdpx seems fine

/Bingo

stephenw10

Technically autonegotiation is mandatory for 1000base-T if it meets the spec for 802.3ab.
https://en.wikipedia.org/wiki/Gigabit_Ethernet#1000BASE-T
Though many devices allow it to be disabled which means you can link to equipment that does not comply with the spec.

Steve

Ghost 0

Hello, I'm currently experiencing similar issue. My WAN keeps dropping every few days. I couldn't figure it out until I read the gateway log that seems to suggest the WAN interface is the issue. Comcast is my ISP, and I own my own cable modem (Netgear CM700). My gateway {pfSense 2.5.2} is an old repurposed HP workstation with a 4-port intel NIC. I have been running pfSense for about two years. I have a fairly complex system with multiple LANs /VLans and a plethora of managed switches. It could be my cable modem that's on its last leg since it has a poor reputation (the controversial PUMA chipset) or my switches due to an improperly configured STP because sometimes the issue would be temporarily resolved after rebooting only the switches. However, the issue continues after properly configuring STP. Refreshing openvpn would also temporarily resolved the issue. Also quickly unplugging and plugging solely the WAN ethernet cable would temporarily resolved the problem. I'm currently at a lost for solutions. It wasn't until today when I finally really read the pfSense logs in details that pointed me to the potential culprit, the WAN interface. Before that, I tried all types of fruitless solutions. I serendipitously stumbled on your potential solution, changing the speed/duplex of the WAN interface from autoselect to 1000 base T full duplex. I have applied it and hope that it works; fingers crossed and will find out in a few days if this was the issue.