I've got a solid setup of demand dialling using diald/pppd. I'm using SUSE
6.3. The connection is using ISDN using an extrenal Hagenuk Speed Viper.
My ISP is having serious problems with :
1) Connections aborting early on. (first few seconds).
2) Connections not authenticating.
3) Connections of Poor quality.
I'm tyring to assist them in helping out.
I've searched various man pages and have configured pppd with the debug
option and kdebug option - so I get everything but the raw packets stored
in logs. I don't think I've got space for the raw packets. :-)
I've also tweaked my setup string for the TA so it is set to report
progress using X2 - which will give me the reason codes for calls that are
not connected.
Examing some duff connections yesterday...
Some calls connect, there then then follows :
Feb 7 20:06:42 lnx1 pppd[3136]: pppd 2.3.10 started by root, uid 0
Feb 7 20:06:42 lnx1 kernel: ppp_ioctl: set dbg flags to 10000
Feb 7 20:06:42 lnx1 kernel: ppp_ioctl: set flags to 10000
Feb 7 20:06:42 lnx1 pppd[3136]: Using interface ppp0
Feb 7 20:06:42 lnx1 pppd[3136]: Connect: ppp0 <--> /dev/ttyS0
Feb 7 20:06:42 lnx1 kernel: ppp_tty_ioctl: set xasyncmap
Feb 7 20:06:42 lnx1 kernel: ppp_tty_ioctl: set xmit asyncmap ffffffff
Feb 7 20:06:42 lnx1 kernel: ppp_ioctl: set flags to 10000
Feb 7 20:06:42 lnx1 kernel: ppp_ioctl: set mru to 5dc
Feb 7 20:06:42 lnx1 kernel: ppp_tty_ioctl: set rcv asyncmap ffffffff
Feb 7 20:06:42 lnx1 pppd[3136]: sent [LCP ConfReq id=0x1
<pcomp> <accomp>]
Feb 7 20:06:42 lnx1 kernel: ppp_ioctl: get unit: 0
Feb 7 20:06:45 lnx1 pppd[3136]: sent [LCP ConfReq id=0x1
<pcomp> <accomp>]
Feb 7 20:06:51 lnx1 last message repeated 2 times
Then it abandons the attempt and the line is dropped.
--------------------
There are three possibilities as to what is going on here
A) The Telco doing the connection haven't actually established a bearer
channel - this would mean that the ISP wouldn't see any of the packets
I've sent and I wouldn't see any reply anyway.
B) The bearer channel between myself and the ISP is being established -
but is very badly corrupted. So no valid packets are being seen
C) There is a problem at the ISP which means it isn't starting the
euthentication process.
To enable me to distinguish between the above - what I really need to know
is if any packets are being received but are being dumped because they
fail the chekcsum/whatever. As far as I can see the Kernel does not log
anything when it chucks a packet away. Is there anyway of doing this - or
of forcing the kernel to at least say how many packets have been dumped
from within a script ?
-----------------------------------
The calls which connect but are then very slow or stall completely have a
different problem.
Looking at the logs shows that the Kernell keeps reporting the following
warning on such connections - and this warning is rare or non-existent
Feb 7 20:07:12 lnx1 kernel: tcp_v4_rebuild_header(): not valid sock
addrs: saddr=00000000 rcv_saddr=00000000
As far as I can tell this problem relates to the rebuilding of headers
which have been VJ compressed. The only hints I get about what is going on
here is from the O'Reilly PPP book which sort-of suggests this can result
from having too many lost packets on a VJ-compressed link. Is this right ?
can I deduce anything else.
What I'd really like is a way to get the kernell to log to a seperate file
any packet which is rejected - so we can try and determine what happened
to it.
Thanks in Anticipation
Andrew
--
To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com
For additional commands send e-mail to suse-linux-e-help@suse.com
Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
