Yes, I noticed that.
Please add a comment to my task saying that you are also experiencing the same pppoe disconnection.
For me DNS is fine.
There is a new patch commited in the Trunk version which should fix my RA problems with TIM. I am going to test it for the next days..so far so good.
Maybe it will fix your DNS problem.
dns still there... how can i underestand the problem ?
Open a ticket with LEDE version, router model and also your NIC driver version (try updating it if you can.. I had problem with IPv6 at work in the past). Attach also your network/firewall/dhcp configuration. Start with the latest trunk version if possible because as I wrote before, it now includes my patch that solved my problem.
Reboot router, reboot your device and start capturing packets.with tcpdump, wireshark.
What to capture:
- enable odhcpd debug level "uci set dhcp.odhcpd.loglevel=7" and save your syslog to your pc
- capture IPv6 only traffic on your device
- capture traffic on the router - interface pppoe-wan or pppoe-wan2 (you know why
) - capture traffic on the router - interface nas0 ( or similar for you router)
I used an USB pendrive attached to my router to save the pcap files due to router space limitation.
I used facebook homepage always opened so it automatically refresh that pages and check if I was still using IPv6 (chrome extension IPvFoo).
Keep track of when the DNS disappears or have problem so it will make troubleshooting much easier.
Wow lots of data...
it's difficult to even reproduce that... will try...
Well, it is a start.
I suppose the developer needs more info.
Attach your /etc/config/network, /etc/config/firewall, /etc/config/dhcp files (sanitize them, removing passwords if any..)
And yes, we need to see tcpdump/wireshark capture to understand more. If you can't capture router's interfaces, as a start you could just attach wireshark pcap from your PC with filter "icmpv6 or dhcpv6". If you leave wireshark opened all day you should have a <100MiB pcap file(remember to 7zip it to attach it).
Do not use pastebin, but attach files directly into the bug report.
Without data it is difficult..
will build an image with tcpdump and will dump on hard drive on modem np... just tell me what to dump... (the exact command would be the best... in the bug report i added my config)
Do a ifdown wan_6 then ifup wan_6 and start capture as soon as it creates pppoe-wan2 interface. This means that when you execute the following command you have to repeat it until it succedes so you won't lose any initial packets ( up arrow + enter, up arrow + enter...)
tcpdump -i pppoe-wan2 -w /path/to/hdd
Ok done... uploading to the bug site
I added a comment in the meanwhile..
I probably shouldn't celebrate yet but I think I fixed the routing issue.
The IPv6 PPP connection was installing a non-working 0.0.0.0/0 via pppoe0 route that overrides the right 0.0.0.0/0 via pppoe1 one. I removed defaultroute from the options for pppoe0 (IPv6) and set replacedefaultroute in the options for pppoe1 (IPv4).
post the network config pls...
also we need the 2 ppp interface anyway
Ok, I don't know how to specifically do this in LEDE but you need to remove defaultroute and add replacedefaultroute to the options of pppd for pppoe0 (the IPv6 PPP connection) and pppoe1(the IPv4 PPP connection) respectively.
In vyatta/edgeos/debian the options are in /etc/ppp/peers/pppoe[n].
These are my two files: pppoe0, pppoe1.
By the way, I was curious last night and I scanned the telecom italia IPv6 prefixes (they have two and are using only one) and the result was pretty amusing: there are only 13 people using IPv6 (including me) and I think I found your router @Ansuel.
ahahahhaha only 13 people how you checked that? (i think 2 of that ip are mine as i currently have another home that have a different connection so.... )
so you modify the ppp option, here in lede we have a configuration file and then the deamon are calle with the option in that config file so we should add that option.... not so diffuclt i think...
anyway the best solution would be to understand why we can't have a normal dual stack connection....
ahahahhaha only 13 people how you checked that?
They seem to only use 2a01:2000:20001::/48 to form a pool of /64 prefixes; assuming each router assign itself [prefix]::1/128 and then does SLAAC (which I think it's reasonable, with a /64 you can't do much else) there is only an hextet of freedom, so 2^16 different prefixes and as many addresses to scan.
I used nmap -sn to do it. This option just sends ICMP echo requests (these shouldn't be blocked by default for IPv6 so it's the best bet) and got 13 hosts up.
With a port scan of these I then found ansuel.com.
i think 2 of that ip are mine
So there's literally a dozen of us!
anyway the best solution would be to understand why we can’t have a normal dual stack connection…
I think this is just the way it's supposed to work. It's not the only ISP doing this: I tried adding a separate connection after reading a thread of a japanese user whose ISP had a similar setup.
then why give a dual stack connection... just give a ipv6 only connection with that specific user and password... i still think there is something wrong with lede or our setup...
Maybe, but I'm now of the opinion that this "test pilot" is more something telecom put together to say they have IPv6 support and show how innovative they are than actually deploying IPv6 to the users.
true.... other isp provide a /48 pool.... can't do much with a /64... anyway do you know some way to know if it's a problem with a misconfigured isp or with some strange configuration on our side?
it would be good to solve it so that we don't need 2 pppoe