[pptp-server] Link fails after increased data flow

[Bruce Morrison]

I'm connecting a linux box to an NT system with out any problems.  I can
transfer data between the networks.  But when the amount of data
increases the link stops responding.

I'm seeing the following 

mppe_decompress0: bad seq # 48, expected 47
ppp0: decomp err -1
ppp0: not decomp, rc_state=c24a2000 flags=f4130cb
ppp0: not decomp, rc_state=c24a2000 flags=f4130cb
ppp0: not decomp, rc_state=c24a2000 flags=f4130cb

and also
pppd[1180]: rcvd [Compressed data] 91 4f f5 d0 c4 a3 83 86 ...
pppd[1180]: rcvd [Compressed data] 91 52 05 01 d1 3c ad 7c ...
pppd[1180]: rcvd [Compressed data] 91 53 c6 00 6a ef c1 7b ...
pppd[1180]: rcvd [Compressed data] 91 54 fc a0 a5 de 2f e4 ...
pppd[1180]: rcvd [Compressed data] 91 55 49 f8 0f fd fd a3 ..

The pptp control channel is still up (I can see the PPTP-ECHO request
and replys).  I can also see GRE packets pasing between the two systems
but pinging the remote end results in failure.  It seems that pppd has
received some bad data and is in a bad state.

Anyone have any pointers ? (Is it that the hardware is just not up to it
?)

The software in use is
SSLeay-0.6.6b 
ppp-2.3.8 (with ppp-2.3.8-mppe-others-norc4_TH7.diff)
pptp-linux-1.0.2
kernel 2.2.17

The hardware is
CPU P133
MEM 48Mb

/etc/ppp/options is

lock
noauth
debug
nodefaultroute

name XXXX\\XXXXX
remotename XXXXX

mppe-40
mppe-128
mppe-stateless

-- 
Bruce Morrison <brucem at sausage.com>
Senior Systems Administrator
Sausage Software, Limited