Negate 6100 - unstable LAN interface
-
Hello,
I recently received 2 brand new Netgate 6100 devices.
I'm experiencing an issue with the LAN1 interface which go down without reason and go up just few seconds later.
I tried to change the network cable connected to my computer through a USB3 ethernet adapter with same result. I changed the USB3 ethernet adapter too with no more luck.
Since I have 2 units, I tried to load the same configuration on the second device but the issue is still there.
I went into the shell on the console to troubleshoot the issue by analyzing logs with no success.
Has someone already seen this behaviour ?
Thank you.
-
I have seen something similar to that once before, it was a link negotiation issue. It only did it when I had the interface link speed specifically set to 'autoselect' rather than default. Is it possible you have set that?
Steve
-
@stephenw10
Thank you for your answer.In the meanwhile I tried to force 100BaseTx FullDuplex to see if the issue persist.
The strange thing is when I set to default again the interface did not switch to 1000BaseT FullDuplex, I had to force autoselect to correctly reset the speed negociation.
For you, the correct setting is "default" ?
-
Yes it should default for almost any connection.
Does it fail to link completely at default?
What is it connected to? Same cable in the tests?
Steve
-
@stephenw10
My computer is connected with a USB3 gigabit ethernet adapter to the LAN1 port of the 6100.I already tested with another CAT5E brand new cable and checked another ethernet adapter with still these interface resets.
-
I set up my LAN interface to default again an restarted the 6100.
Is there any tools/logs to troubleshoot link negotiation errors ?
-
When I saw this it was cycling at ~5s, the system logs will show the timing.
What happens when you set the speed to 'Default'? Does it link at all?
One thing you can try is setting both ends the 1G fixed but that shouldn't be required for Gigabit Ethernet.
Steve
-
With "default" parameter I can access the web admin but sometimes the interface resets.
I put an unmanaged cisco smb gigabit switch between the netgate and my computer, it looks promising.
-
Hmm, almost certainly a negotiation issue then.
-
Test with this :
Take a 100Mbit or better 1Gbit ordinary switch.
Hook up pfSense port LAN1 to the a port on the Switch.
Hook up the Ethernet cable coming from your USB dongle to the same switch.
I bet the issue is gone.
And if it persists, it's on the "USB dongle side" of the switch port, not the pfSense port.Because, look closely to the USB electrical contacts. The traffic has to go over these 4 contacts. Not some xx kbit/sec, but at least 100 Mbits/sec (GBits is even possible ?)
and
The needed power also goes over the same contacts. Knowing that Ethernet type device doesn't use "one or two millli amps" but a lot more ... If the USB dongle even vibrates a bit, power to the USB NIC goes flat.
The entire Ethernet links goes down, and up again.IMHO : USB Ethernet dongles are usable at best as an extreme "no other options left" Ethernet connection solution. Even a wifi USB dongle is better ( the connection still breaks, but you won't see it any more so it "doesn't happen" ;) )
-
As I said I tried this configuration with a switch and can now confirm the issue is gone !
The 6100 is now installed, thanks for your help @Gertjan @stephenw10
-
@stephenw10 said in Negate 6100 - unstable LAN interface:
I have seen something similar to that once before, it was a link negotiation issue. It only did it when I had the interface link speed specifically set to 'autoselect' rather than default. Is it possible you have set that?
Steve
When setting up my 6100, I had this issue when I chose AutoSelect. I changed it back to Default and the issue went away.
-
I appear to have had this problem as well.
For my 6100, if you do a full reboot, everything is stable for a long time, but if the link goes down (switch reboot, or unplug/replug cable), the LAN port would go into an endless cycle of up/down.
I discovered that the LAN port was set to "autoselect" which must be the default, as I did not set that at all. When set to "default" it seems to be working now (for now).
My questions are:
- is this a hardware issue or a software bug?
- what is the difference between "default" and "autoselect" (and why the difference?)
- how is this not a widely known issue? TAC had no clue when I contacted them right after I got the unit.
thanks,
-Randy -
Setting 'autoselect' specifically sets the NIC to auto negotiate speed and duplex. Setting 'default' doesn't set anything so the NIC remains set at it's power on values. That usually is 'autoselect' but may be slightly different for some of the sub settings.
After booting when you set 'default' it may not change what was previously set. So for example if you have set 100Mbps and want to then connect at 1G you may need to set autoselect first then default.
Does the link remain stable with the NIC set as 'Default'?
Steve
-
After setting it to "default" (which oddly, seems not to be the default), it does seem to be stable.
I have tested this on a 2.5G and 10G port on the LAN side and both seem to function just fine in terms of link. Getting the unit to properly re-assign IPv6 addressing on the LAN side is a different story, but a release/renew on the WAN fixes that.
I am still a bit confused as to why there would be a difference. All 1G and 10G should be set to auto-negotiate, should they not? Is there any documentation as to exactly what "default" does at the interface level?
thanks,
-Randy -
Setting 'autoselect' there runs 'ifconfig media autoselect' against the NIC which may also set/unset some other registers in the NIC. Setting 'default' does not.
