[29] in linux-net channel archive
MRU problems in 1.1.83 PPP or am I dreaming?
daemon@ATHENA.MIT.EDU (G.W. Wettstein)
Mon Jan 23 19:35:05 1995
From: wind!greg@plains.nodak.edu (G.W. Wettstein)
Date: Mon, 23 Jan 1995 16:13:53 CST
To: linux-net@vger.rutgers.edu
Cc: linux-activists@niksula.hut.fi
Hopefully this isn't something utterly stupid on my end. Due to my
relative lack of expertise with IP over serial this could well be the
case.
We needed to start providing PPP service to physicians working from
home. As a result I spent the weekend becoming intimately familiar
with this component of networking. First of all I would like to thank
everyone involved with bringing PPP to Linux. I was preparing myself
for hours of frustration. Instead my experience was compile the pppd
daemon, read the docs and go. Operation has been predominently
flawless.
The only catch that I seemed to run into was when I lowered the MRU
(MTU?) to 296. The default is 1500 which makes interactive work
slightly bawky. Following the instructions in SETUP I added an mru
296 to the options file on both ends of the length.
Immediately we began to have problems with PPP failing to negotiate a
link between the client and the peer. The two PPP daemons would start
talking but endlessly fail the negotiation. I deleted the debugging
files before I decided that I should probably send a note to the list
about the event. The peer was sending negotiation frames and the
client was complaining about receiving LCP frames when it wasn't
supposed to or perhaps receiving other frames when it was expecting
LCP frames. If need be I can generate the errors again and forward
them to whoever is interested.
The two daemons were essentially deadlocked. I let them run until
they generated about 0.5 megabytes of debugging information and they
never synchronized. At first I thought it was problem with
authentication but I switched that off and the difficulty persisted.
The two daemons would properly negotiate a connection in about 1 of 20
attempts.
I finally commented out the mru 296 on each side of the link and the
problem immediately vanished. We are using CHAP authorization and
have not yet had a single negotiation failure.
Both sides of the link were using the 1.1.83 kernel and the pppd
daemon was compiled with 2.5.8 of gcc with the 1.1.83 header files.
The source I used was from the ppp-2.1.2b package on sunsite.
Has anyone else seen this or is this gross operator error? I would
like to lower the MRU but cannot risk having machines deadlock with
inexperienced user's in the field. Any comments welcome, if someone
needs more debugging information please feel free to drop me a note.
Have a pleasant week everyone.
As always,
Dr. G.W. Wettstein Oncology Research Div. Computing Facility
Roger Maris Cancer Center UUCP: uunet!plains!wind!greg
Fargo, ND 58122 INTERNET: greg%wind.UUCP@plains.nodak.edu
Phone: 701-234-2833
----------------------------------------------------------------------
`The truest mark of a man's wisdom is his ability to listen to other
men expound their wisdom.' -- GWW
