[solved] PPPoE IA-PD woes

maverick_slo

Ummm see post before last one.. There are logs what happens on cable unplug…
Tried patch aswell same results...

Guest

And that is not what the log should say if you have the correct patches in place.

Mar 11 21:04:55 rtsold RTSOLD Lock in place

has been replaced.

The current debug message says "RTSOLD Lock in place - sending SIGHUP to dhcp6c.". I changed it so I could be sure testers were running the correct patches.

The patches are:

718cbc2d3921627e9767e59d539386c843dffcc4
ea4de27b15441763bd227a2a942df522d98775c8
2d81d94e1d96dacae82cf4c60a4f069c066d22ec

Applied in that order.

And one more to add, just found another posix_kill() that would leave the lock behind.

e660bd4f34245958a290240c04afacb0823d4b74

maverick_slo

Hmm yes I know it changed to RTSOLD Lock in place - sending SIGHUP to dhcp6c. But patch unfortunatley has no effect. Will try latest tomorrow and report back.

maverick_slo

Hehe now you nailed it :)
It works perfect with one exception which I dont know if its related…

Tried 10 times.
6 times all ok but in 4 times I got no IPv6 address on my pppoe iface.
Everything works as expected exept gateway monitoring (it shows status pending because ipv6 address is missing on ppoe).

I don`t know why it happens it seems really random.

Thanks!

maverick_slo

Ha! Managed to replicate when pppoe does not obtain ipv6 address…

So if link flaps for 10 seconds or just modem reboots there is no problem.

But when link is down more than that and LCP echo requests start to fail or pppoe reconnect after 5th LCP missing is exectuted, then pppoe does NOT obtain IPv6 address and gateway is marked as pending.
IPv6 from firewall and subnets is working just fine, again just gateway monitoring is impacted...

BR,
Greg

Guest

That's good news. My bad for not thouroughly checking for any other posix_Kill() calls for dhcp6c so sorry for that.

I have noticed that sometimes the gateway monitor does not reflect the real state after an up/down event. I tend to no believe it if I can see the addresses are all there and just leave it, the page refresh always seems to sort it out after a while.

I'll merge those commits into one later today and update the PR.

doktornotor

@marjohn56:

The patches are:

718cbc2d3921627e9767e59d539386c843dffcc4
ea4de27b15441763bd227a2a942df522d98775c8
2d81d94e1d96dacae82cf4c60a4f069c066d22ec
e660bd4f34245958a290240c04afacb0823d4b74

At the risk of looking dense - what patches where? Is there some PR for this?

maverick_slo

I would be sarcastic but Ill leave that to you good ol doc :)

https://github.com/pfsense/pfsense/pull/3515

Guest

@maverick_slo:

I would be sarcastic but Ill leave that to you good ol doc :)

https://github.com/pfsense/pfsense/pull/3515

:D

Note that if you apply the lot you'll get REASONS working too, you should see some effects from dhcp6 renews no longer triggering wan updates, makes the logs much quieter.

doktornotor

I did not mean to be sarcastic, I wanted to make sure that there's a proper PR for this, otherwise you end up with half of the patches getting lost and things broken

maverick_slo

J/K :)

About my IPv6 on pppoe iface issue, should I start new thread/redmine or is it somehow connected to this?

Guest

@maverick_slo:

J/K :)

About my IPv6 on pppoe iface issue, should I start new thread/redmine or is it somehow connected to this?

It's ALL related in there somewhere, but gateway monitoring is not an area of pfsense I'm familiar with, so whether its related to locks and reasons, i'm summing the two, or something else thats always been there I have no idea… at the moment.

Guest

I'll snapshot my test system and do the patches, then I'll throw some curved balls at it and see if I can replicate the monitoring issue. On a side note, I'm changing ISP in a couple of weeks and my v6 will no longer be dhcp6 without RA ( that's a mismomer and we should change it! ) but similar to Mav*Slo, so I will be able to see exactly what's going on there.

maverick_slo

No no, gateway monitoring is just consequence…

Look what I mean:

When things are 100% OK:


Status
    up
PPPoE
    up 
Uptime
    00:01:50
MAC Address
    00:00:00:00:00:00
IPv4 Address
    212.xx.xx.xxx
Subnet mask IPv4
    255.255.255.255
Gateway IPv4
    212.xx.xx.xxx
IPv6 Link Local
    fe80::xxx:xxx:xxxx:xxxx%pppoe0
IPv6 Address
    2001:xxxx:xxx:x:xxx:xxx:xxxx:xxxx
Subnet mask IPv6
    64
Gateway IPv6
    fe80::xxx:xxxx:xxxx:xxxx
MTU
    1492
In/out packets
    1543787/651379 (1.61 GiB/46.27 MiB)
In/out packets (pass)
    1543787/651379 (1.61 GiB/46.27 MiB)
In/out packets (block)
    2393/0 (137 KiB/0 B)
In/out errors
    0/0
Collisions
    0

And when there is no IPv6 on PPPoe:


Status
    up
PPPoE
    up 
Uptime
    00:00:23
MAC Address
    00:00:00:00:00:00
IPv4 Address
    212.xx.xx.xxx
Subnet mask IPv4
    255.255.255.255
Gateway IPv4
    212.xx.xx.xxx
IPv6 Link Local
    fe80::xxx:xxx:xxxx:xxxx%pppoe0
Gateway IPv6
    fe80::xxx:xxxx:xxxx:xxxx
MTU
    1492
In/out packets
    1545272/656932 (1.61 GiB/46.65 MiB)
In/out packets (pass)
    1545272/656932 (1.61 GiB/46.65 MiB)
In/out packets (block)
    2418/0 (137 KiB/0 B)
In/out errors
    0/0
Collisions
    0

Guest

Can you do a test for me, go to the shell, and firstly check that dhcp6c is running, if it is, then can you send it a SIGHUP, SIGHUP being 'kill -1 PID', see if that makes it recover.

maverick_slo

DHCP6c IS running.

root    39375   0.9  0.1   8348   2328  -  Ss   10:04      0:00.00 /usr/local/sbin/dhcp6c -D -c /var/etc/dhcp6c_opt2.conf -p /var/run/dhcp6c_pppoe0.pid pppoe0

Killed it:


kill -1 39375

No change on pppoe :(

maverick_slo

And quite a lot of:


Mar 12 10:14:33 	php-fpm 	80662 	/rc.newwanip: ROUTING: setting IPv6 default route to fe80::xxx:xxx:xxxx:xxxx%pppoe0
Mar 12 10:14:33 	php-fpm 	60486 	/rc.newwanipv6: rc.newwanipv6: No IPv6 address found for interface PPPOE [opt2].
Mar 12 10:14:33 	php-fpm 	60486 	/rc.newwanipv6: rc.newwanipv6: Info: starting on pppoe0\. 
Mar 12 10:14:31 	php-fpm 	80662 	/rc.newwanipv6: rc.newwanipv6: No IPv6 address found for interface PPPOE [opt2].
Mar 12 10:14:31 	php-fpm 	80662 	/rc.newwanipv6: rc.newwanipv6: Info: starting on pppoe0.
Mar 12 10:14:31 	rtsold 		Starting dhcp6 client for interface opt2(pppoe0)
Mar 12 10:14:31 	rtsold 		Received RA specifying route fe80::xxx:xxx:xxxx:xxxx for interface opt2(pppoe0)

Guest

What do the dhcp logs show at that moment?

maverick_slo

Actually same as when it works… I see nothing special there...

Guest

So it appears dhcp6c is working, and by the look of it so is RTSOLD. Hmm, pppoe thing. Not even sure if it's related or not.

Try this, see if it makes any difference. Just add it on the end of the others.

c3b55b8a0aee8fde7474738d08a993ed32ee3282