topic Re: Intermittent IPv6 connectivity in IPv6 Trial

Intermittent IPv6 connectivity

wmh — Thu, 28 May 2026 10:27:39 GMT

My IPv6 connectivity is dropping out each time the PPPoE connection drops or is reset. To be clear, IPv6 does work but I am regularly seeing the following faults.

I see two different symptoms:

1. Sometimes the /56 prefix is not allocated via DHCPv6, and the router tells me that all of the IPv6 pools are missing. Simply disabling, waiting 5 seconds and then re-enabling the PPPoE interface is usually sufficient to fix this. This has happened 3 times now.

2. Almost every time the PPPoE connection resets and comes back up (and the pools allocate correctly), IPv6 still does not work. This has happened lots of times now. In this state I see:

- IPv6 addresses allocated to the clients correctly

- IPv6 traffic showing on the firewall connections monitoring tab

- TCP connections showing "syn-sent". But never "established".

- IPv6 web test page shows "IPv4 only".

- Monitoring (Torch) on the PPPoE shows outbound packets but never any replies.

Sometimes disabling and re-enabling the PPPoE fixes it, quite often not. Sometimes disabling the "add default route" and then disabling, reenabling the PPPoE and then finally reenabling "add default route" fixes it. Sometimes not. Repeating these, and eventually it comes back up and "just works". But if left alone, it appears to stay broken.

Once I have got the Plusnet IPv6 working, it appears to stay up, no issues, until the next time the WAN link drops or resets.

Some additional information that may be relevant:

Current hardware: Mikrotik RB5009. Fully up to date on the firmware. I cannot see anything being blocked or denied on the WAN link. I have added the "clamp to pmtu" fix noted in this forum (no change, I dont think I was suffering this issue anyway).

I also suffered from the same loss of connectivity on the previous UniFi UCG Ultra, including after being factory reset. So I dont think this is Mikrotik specific. Or configuration specific seems how it also fails "out of the box".

I also have a HE IPv6 tunnel that I need to keep for various reasons. It never drops. (I am using some source routing to direct the subnet to the tunnel. I do not mix plusnet or HE IPv6 on the same VLAN.).

IPv4 connectivity never drops. That appears as rock solid as always.

Any suggestions or advice from anyone?

Re: Intermittent IPv6 connectivity

MPC — Thu, 28 May 2026 11:07:02 GMT

I haven't seen this, but the number of times my wan goes down is counted on one hand per year.

Regardless, I use a fairly simple bash script on my router that checks I can ping a reliable IP6 address every minute, and if it fails, I log to ntfy.

If it fails more than 3 times, I stop and restart wide-dhcpv6-client and log to ntfy, then wait 10 minutes before resuming per-minute checks. I do the same for the IP4 side.

(network_vars.sh includes network_vars_plus6.sh which is written as part of ppp's ip-up processing to log network info. You could pick any reliable ip6 host - I use google.dns IP6 address )

#!/bin/bash

# Script to monitor the stability of the plusnet IP6 connection

COUNT=0

TIMEOUT=3

PINGS=2

PINGWAIT=7

INTERTEST=60

POSTDOWNUP=600

function cleanup()

{

echo NOTICE: Exiting script

exit 0

}

trap cleanup SIGINT

NOW=$(date)

echo "[$NOW] NOTICE: monitor_plus6 $$ started"

echo

while true;

# These may change while we are monitoring so load them now

. ~/bin/network_vars.sh

TESTIP=$VALID_PUB_GW6_PLUS

# Try pinging gateway $PINGS times or for $PINGWAIT seconds

ping6 -c $PINGS -i 2 -w $PINGWAIT "$TESTIP" 2>&1 >/dev/null

if [ $? == 0 ];

then

if [ $COUNT -gt 0 ];

then

NOW=$(date)

echo "[$NOW] NOTICE: ping6 test successful after $COUNT failures"

echo ==============================================

# If all ok then reset the counter and wait before retrying

COUNT=0

sleep $INTERTEST

continue

# Increment the ping check count

let COUNT=$COUNT+1

NOW=$(date)

/usr/bin/curl -m 5 -u "ntfypublish:<censored>"

if [ $COUNT -lt $TIMEOUT ];

then

echo "[$NOW] WARNING: ping test to $TESTIP failed $COUNT times"

sleep $INTERTEST

continue

# Ping has failed every $PINGWAIT for $COUNT times. Down and up the interface

/usr/bin/curl -m 5 -u "ntfypublish:<censored>"

echo ======================================================================

echo "[$NOW] ERROR: ping test to $TESTIP failed $COUNT times"

echo "[$NOW] ERROR: wide-dhcpv6 stop and start is next"

echo ======================================================================

trap - SIGINT

systemctl stop wide-dhcpv6-client.service

sleep 5

systemctl start wide-dhcpv6-client.service

trap cleanup SIGINT

NOW=$(date)

echo ======================================================================

echo "[$NOW] ERROR: stop - start completed. Waiting $POSTDOWNUP before testing again."

echo ======================================================================

# Wait before trying again afer downing and restoring the interface

sleep $POSTDOWNUP

done

Re: Intermittent IPv6 connectivity

wmh — Fri, 05 Jun 2026 07:41:35 GMT

As I have mentioned in the other thread, it dropped again last night but came back up on its own this time.

However, it did come back up again with a new prefix. Everything has picked up new IP addresses this morning. (The previous outages and general bouncing everything around did not change prefix, including changing the router)

Re: Intermittent IPv6 connectivity

dave — Fri, 05 Jun 2026 08:04:53 GMT

Sorry, should've replied to you earlier, I'm starting to move the process to the "go live" allocation of IPv6 prefixes. At the moment for the trial the IPv6 prefix is assigned to your account but the place that it's stored in the database is the same place we use for the Broadband Firewall and I've seen a couple of addresses get removed when the Broadband Firewall setting is changed.

That's obviously not a permanent solution, so I'm creating a new table where the IPv6 address is associated with the IPv4 address. In doing so I noticed that 1 IPv6, your address, was incorrectly stored in that database, I think that's why you were getting intermittent issues because it wasn't allocating it properly, so I ran a fix last night and in doing so looks like it switched you for the next range up. Hadn't realised it was going to do that. It should though be permanent now.

Re: Intermittent IPv6 connectivity

wmh — Fri, 05 Jun 2026 09:17:18 GMT

Thank you!