FTTC PPPoE drops every 5 days

[#1]

Seeing regular dropouts in PPPoE every 5 days at same time of day.

setup: FTTC -> Master socket -> BT Openreach box -> Ubiquiti EdgerouterX 

Timing of one of the log messages from the ERX as it reconnects shows is below. 

As you can see it's happening very regularly - pretty much exactly every 5 days when accounting for the reconnect time.  The ERX is new, but I was seeing similar behaviour and frequency with a Unifi USG, although I never got to the point of logging into console to confirm exact timings, but it tended to be shortly after 8PM GMT when it happened until we turned the year into 2021 when it started kicking in shortly after 9PM!!?? 

Don't need to take any action, just wait a few mins then it sorts itself out, but its frequently in the middle of presenting in work calls which is very inconvenient.

In terms of potential local interference from other devices in the house there's nothing I run or am aware of that kicks in on that 5-day frequency.  Also really odd that it shifted from around 20:00 to 21:00 from Jan 1st!

A real puzzle - interested if anyone else is seeing anything similar and what can be done to resolve.

Thanks!  JR

ubnt@EdgeRouter-X-5-Port:/var/log$ grep CHAP messages
Jan 4 21:04:21 EdgeRouter-X-5-Port pppd[3000]: CHAP authentication succeeded
Jan 9 21:07:30 EdgeRouter-X-5-Port pppd[3000]: CHAP authentication succeeded
Jan 14 21:08:54 EdgeRouter-X-5-Port pppd[3000]: CHAP authentication succeeded

[#2]

Can you think of anything that powers on / off at that time? I've heard of street lamps causing weird internet issues in the past...sounds like something on a timer doesn't it?

Could a neighbour have something on a timer?

[#3]

I have regular (well, not actually all that 'regular') line resyncs, time can vary from a week to a couple of hours (but typically 2 to 5 days).

I have opened a case with PlusNet, so my Hub1 is currently connected to the "test socket" (all my phones are connected via a central filter, so it made no real difference), and I am logging all resyncs and rests.

Strange that yours are amazingly every 5 days, to within a few minutes. Keep logging the information, it might help you / plusnet work out what's going on!

[#4]

Thanks orby - agree - there's definitely something on a timer or schedule - but I'm convinced there's nothing in my property and I've asked the immediate neighbours if they've anything working on that schedule and all negative.  I've read something similar about street lamps but I've ruled that out because of the time of year.  Almost feels like its some sort of maintenance kicking in upstream.  I'll keep pondering...

Cheers

[#5]

Ta Paul - I'll log a formal case with them too and will probably end up plugged into the test socket like you shortly.  Will keep gathering the logs and watching for patterns.

Its the sort of thing that were it happening at 2 or 3AM I'd never have noticed!  But its just at a really inconvenient time when I really need a stable connection.

JR

[#6]

@whoshotjr  are there any LCP timeouts in the ERX log ?

This https://community.ui.com/questions/ER-X-ADSL-PPPoE-resets-from-PPP-LCP-timeouts/cedee82d-f89e-4f5b-8... might be worth a read

[#7]

Interesting that I'm having similar problems with fairly regular dropouts at about the same time each evening and I'm certain they are caused by something external to my house.

All the Plusnet checks make no difference.  An immediate neighbour (with EE) has similar experiences.

I don't really understand the router Event Log, but at the time of the dropout, it usually shows something like:-

20:36:14, 18 Jan.

(2029043.380000) PPP LCP Send Termination Request [User request]

To me, this implies that I terminated the connection - if so, this is not true.

Does anyone know what's going on and what to do about it ?

Thanks.

[#8]

To me, this implies that I terminated the connection - if so, this is not true.

You didn't but your router did. Is there anything in the log prior to that ? such as LCP timeouts or DSL dropping

[#9]

With apologies to the OP, didn’t intend to hijack your thread. 
In reply to MisterW, my Router Log has no mention of LCP or DSL prior to the dropout. These are the relevant entries - if they make sense to you :-

20:36:23, 18 Jan.	(2029051.910000) CWMP: Session start now. Event code(s): '4 VALUE CHANGE'
20:36:23, 18 Jan.	(2029051.630000) CWMP: Initializing transaction for event code 4 VALUE CHANGE
20:36:19, 18 Jan.	(2029047.970000) PTM over DSL is down after 1397 minutes uptime
20:36:19, 18 Jan.	(2029047.960000) PPPoE is down after 1397 minutes uptime [Waiting for Underlying Connection (WAN Ethernet 7 -​ Down)]
20:36:14, 18 Jan.	(2029043.380000) PPP LCP Send Termination Request [User request]
20:33:44, 18 Jan.	(2028893.110000) Host Steves-​iPhone (MAC ca:aa:52:14:d4:16) using IP 192.168.1.74 detected on interface ath10, port -​1
20:29:53, 18 Jan.	(2028661.660000) Lease for IP 192.168.1.70 renewed by host COM-​MID1 (MAC d8:61:62:44:b1:1f). Lease duration: 1440 min
20:29:53, 18 Jan.	(2028661.660000) Device connected: Hostname: COM-​MID1 IP: 192.168.1.70 MAC: d8:61:62:44:b1:1f Lease time: 1440 min. Link rate: 117.6 Mbps
20:29:53, 18 Jan.	(2028661.590000) Lease requested

[#10]

@2000sj  the DSL link does drop

2029047.970000) PTM over DSL is down after 1397 minutes uptime

Ok so that's reported a few seconds after the PPPoE terminate. What I suspect actually happened is the DSL dropped , the PPPoE saw it ( probably no LCP response ), disconnected and reported first , followed by the PTM link over DSL reported as dropped

[#11]

@MisterW - sorry for the long delay in responding on my thread - "life" got in the way.

Think you may well be onto something with that link - there is some interesting detail in my ERX's ppp_pppoe0.log that do indicate its timing out - with the magic number of 6 attempts from the config file mentioned in the thread:

"No response to 6 echo-requests"

And, I was on a USG before and they both are Vyatta-based so that explains why I saw the same behaviour before I moved to the ERX.

I will have a play at the weekend and update the default template to double the failed attempts and see if that has an impact.

So it'll be something upstream of the ERX like the OpenReach modem, or a component in the fibre cabinet that's not responding reliably to these echo requests?

Many thanks for engaging and pointing me in this direction.