WAN fails to keep DHCP address on cable modem reboot
-
@BlankSpace said in WAN fails to keep DHCP address on cable modem reboot:
I believe it is already set to automatic by default.
these are usually automatic by default...
your two devices (modem / pfSense eth.ports) probably there are different - different eth. controllers (Intel, Realtek, etc.)
these controllers often have a trial with speed matching - especially Realtek stuff and thus are placed in a speed matching loop
-
I forced the speed at 1000 and duplex at full, bounced the WAN interface and it came back up and grabbed the IP with no issues. This may be the solution...
-
Nice. You still get the expected throughput?
-
Well, I am not sure. I thought it looked like my RTT and RTTsd increased a little and then I did some speedtests from the pfs command line and they were all over the place. So it is possible my throughput suffered, or just busy cable at this time of the night. I will need to try some speedtests at different times to confirm.
However, since it seemed to solve the problem by adjusting the negotiation, what other workarounds would there be if indeed my throughput suffered from forcing negotiation?
-
If you set a fixed speed at one end you would usually have to set it fixed at both ends. Otherwise the other side might default to 100Mb half or something.
Putting a switch in between would also remove negotiation issues between those two specific devices as both then negotiate with the switch instead.Steve
-
Ah, yes I forgot about that option. Any reason it only happens if the interface is disabled or loses the link with the cable modem when the modem reboots? When I restart pfs there are no issues.
-
Hmm, well that might imply it negotiates differently when the modem comes up.
If your modem really is just a modem and doesn't have a switch IC then I would expect to see a speed/duplex mismatch if you set pfSense to fixed. That usually kills throughput as you get loads or errors on the interface.
Adding a switch in between would be a better test in that situation.
Steve
-
I had two modems that did this, the first was the cable company provided Arris CM8200. The Arris modems would actually go into a boot loop everytime pfs dropped the IP and tried to renegotiate. They would actually lose signal on the cable end and just keep looping until pfs was restarted.
Then I purchased my own Netgear Nighthawk CM1200. The Netgear has 4 ports but it is not a switch per se, it is for link aggregation and multiple IP address.
I just tested speed this morning when manually forcing 1000/full and it did indeed go down, by alot, each time I did the speedtest.
I do have some spare switches laying around that I will try. It's not really too big of an issue because I can always restart pfs and in 60 seconds its back up anyway. But sometimes I want to quickly bounce the cable modem to clear the logs and not have to restart pfs.
-
It would confirm it's a link problem of some type though. Adding a switch is a good test a least.
Steve
-
@BlankSpace said in WAN fails to keep DHCP address on cable modem reboot:
This may be the solution...
Due to the plenty of "noname" chip (PHY, eth. magnetics, PCB. etc.) manufacturers, unfortunately we encounter more and more of this problem.(Free after... @chpalmer https://www.badmodems.com/)
(the bad principle lately is that the modem should be cheap and not really the function is the point)
It's good if your test was successful.
you may still be able to refine your settings with these parameters (Protocol timing)
-
Thanks, I will need to read up on those settings, I have never messed with them before and do not know what each does.
-
@BlankSpace said in WAN fails to keep DHCP address on cable modem reboot:
Thanks, I will need to read up on those settings,
Okéééé:
https://docs.netgate.com/pfsense/en/latest/book/interfaces/ipv4-wan-types.html -
So I put a new cheap unmanaged gigabit switch (Netgear GS205) in between pfs and the modem. And it does the same thing, pfs can't get a stable link, keeps dropping and connecting in a loop. If I restart pfs, no issues.
During this time, this is being displayed on the console: "config_aqm Unable to configure flowset, flowset busy!"
Can it be related to the limiters I have created?
-
Yes. It could be.
Though that could also just be a symptom of the limiter queues filling when the link goes down.Try removing them.
You might also disabling flow control on igb3 if it is enabled.
Steve
-
I'm basically running upload and download limiters using FQ_CODEL for bufferbloat. I just followed the documentation in a video I found.
igb3 is my WAN interface.
-
@stephenw10 "You might also disabling flow control on igb3 if it is enabled."
in addition I would even disable EEE (in loader.conf.local)
-
Thanks, I created loader.conf.local in /boot and added those lines. Safe to assume if I only have igb0 through igb3, that's all I need?
-
Yes. It won't hurt adding more but it won't do anything except maybe log an error since you only have 4 NICs, those further OIDs don't exist.
-
@BlankSpace said in WAN fails to keep DHCP address on cable modem reboot:
only have igb0 through igb3, that's all I need?
As @stephenw10 says too
"prtsc" shows in my example one I350-T4 and one I350-F4 are configured so it shows 0 to 7.
