Hi there,

FWIW my ADSL has been down for 5 hours with a similar error to those reported today.

I've noticed a bunch of posts about ADSL lately and so I've put together a few 
observations about ADSL debugging. I thought they
might help those of you having troubles with your link. Comments welcome as its
very incomplete. Maybe we can get a decent ADSL-Australia-Debug-HowTo out of it all?

I use pppoe from the Roaring Penguin guys. It's great. Its connected to
an ethernet port (eth1 in my case)

1) If you want to see if your connection's concentrator:

eurekait>pppoe -i eth1 -A
(ie eth1 is the ethernet port your modem is connected to, yours may be eth0, etc)

which returns:

Access-Concentrator: nkw2-kent
AC-Ethernet-Address: 00:03:42:15:80:02
--------------------------------------------------

All this proves to my thinking is that the phone line is in place and
works as far as the exchange, your modem is up, talking through the
ethernet port and can communicate, to some degree, to the Access
Concentrator.

This is a good start..

2) This will tell you what's been happening in the syslog re: ADSL Modem
Hangups.

eurekait>echo 'This shows the various connect times for the ADSL line
based on hangups';grep 'Connect time' /var/log/syslog

Heres the last bit of mine:
Jan 30 01:41:41 eurekait pppd[13169]: Connect time 17.2 minutes.
Jan 30 02:52:00 eurekait pppd[13169]: Connect time 70.4 minutes.
Jan 30 05:26:18 eurekait pppd[13169]: Connect time 154.3 minutes.
Jan 30 07:23:01 eurekait pppd[13169]: Connect time 116.7 minutes.

Last hangup was at 7:23 my time. As you can see, the line dropped a lot last night.

3) If you're having issues connecting, this will probably be in your
syslog.

eurekait> tail -f /var/log/syslog

Jan 30 10:30:26 eurekait pppd[5568]: Connect: ppp0 <--> /dev/pts/0
Jan 30 10:30:26 eurekait /etc/hotplug/net.agent: assuming ppp0 is
already up
Jan 30 10:30:57 eurekait pppd[5568]: LCP: timeout sending
Config-Requests 
Jan 30 10:30:57 eurekait pppd[5568]: Connection terminated.


>From what I can tell, having chatted to the guys at the other end, this
requires a 'physical reset' at the Telstra end.. YKMV (Your Kilometers
May Vary)

4) Debugging pppoe the verbose way. 

Built into pppoe is a script that can be altered to provide TONS of
debug information. In my ugly hack of the script I insert a line to
indicate that I require DEBUG.

eurekait>which adsl-start
/usr/sbin/adsl-start

eurekait>vi /usr/sbin/adsl-start
I
<scroll down to a blank line>DEBUG="1"<RETURN>
ESC
:wq
eurekait>adsl-start

This will leave a debug directory in your /tmp/
for example :

/tmp/pppoe-debug-30410/

This file will tell all:   pppoe-debug.txt

An example: Mine.

..skip lots of config data.... useful to check but boring...

Using interface ppp0
Connect: ppp0 <--> /dev/pts/15
sent [LCP ConfReq id=0x9 <mru 1492> <magic 0xe8464a70>]
sent [LCP ConfReq id=0x9 <mru 1492> <magic 0xe8464a70>]
PADS: System-Error: No more sessions allowed for this host
Script /usr/sbin/pppoe -p /var/run/-adsl.pid.pppoe -I eth1 -T 40 -U  -m
1412    -D /tmp/pppoe-debug-30410/pppoe-debug.txt-0 finished (pid
30745), status = 0x100Modem hangup
Connection terminated.
using channel 239

.. skip lots more debug stuff...

This line>PADS: System-Error: No more sessions allowed for this host
Indicates that the exchange thinks I'm already connected, which is not
really possible since I know I'm not????

Once the ISP guy heard this he contacted Telstra to 'reset' the connection
(Don't ask me).


5) The old firewall trick.

pppoe can be made to startup a firewall script once connected. ie there
are options NONE, STANDALONE, MASQUERADE. These point to scripts for
shorewall firewall stuff (on Mandrake anyway). I prefer to use NONE and
have a seperate IPTABLES firewall script that is always up. One chap who
rang me for help, this was the problem, the firewall was blocking some
of the required communication. Make sure of your firewall rules and
where they are being ran from. The Mandrake defaults are pretty good but
there are better out there.

Example from /var/log/syslog

Jan 22 09:25:23 morticia kernel: Shorewall:OUTPUT:REJECT:IN= OUT=ppp0
SRC=144.137.111.174 DST=61.9.192.14 LEN=85 TOS=0x00 PREC=0x00 TTL=64
ID=33579 DF PROTO=UDP SPT=1029 DPT=53 LEN=65 
Jan 22 09:25:23 morticia kernel: Shorewall:OUTPUT:REJECT:IN= OUT=ppp0
SRC=144.137.111.174 DST=61.9.192.15 LEN=85 TOS=0x00 PREC=0x00 TTL=64
ID=33579 DF PROTO=UDP SPT=1029 DPT=53 LEN=65 

I think that the connection was fine but to the user, it looked like the
link wasn't working.


6) Your /etc/resolv.conf

When PPPOE starts up it call pppd which, if it establishes, adds lines
to the resolv.conf to indicate the nameservers send to it from the ISP.
Sometimes this seems to cause name resolution problems, don't ask me why
but deleting the duplicated lines causes the problem to go.

eurekait>cat /etc/resolv.conf

search com dsl.xxxxxx.net.au
nameserver 127.0.0.1
nameserver 61.8.0.113
nameserver 210.23.129.34
# ppp temp entry
nameserver 61.8.0.113 # ppp temp entry
nameserver 210.23.129.34 # ppp temp entry
nameserver 61.8.0.113 # ppp temp entry
nameserver 210.23.129.34 # ppp temp entry
nameserver 61.8.0.113 # ppp temp entry
nameserver 210.23.129.34 # ppp temp entry
nameserver 61.8.0.113 # ppp temp entry
nameserver 210.23.129.34 # ppp temp entry
nameserver 61.8.0.113 # ppp temp entry
nameserver 210.23.129.34 # ppp temp entry
yadda yadda yadda....

See what I mean. Not that this should really upset name resolving but it
does. 


Guess that's it, 

HTH


Stu

-- 
SLUG - Sydney Linux User's Group - http://slug.org.au/
More Info: http://lists.slug.org.au/listinfo/slug

Reply via email to