Re: Strange crashes ...|
Damion Wilson <dwilson,AT,ibl,DOT,bm>|
Fri, 1 Nov 2002 15:22:54 +0100|
I don't know if MTU differences could cause this. CIPE doesn't have to get
into packet reassembly/splitting because that's TCP/IP's job. Have you tested
things with the peers on the same LAN ?
On Thursday 31 October 2002 06:19 pm, you wrote:
> HI Damion,
> had a little time of more testing about that strange behavior ...
> It looks that I have sent you a little bit into a to far direction: pre14
> does also crash in this test environment, but it takes MUCH longer, until
> it crashes the tunnnel and subsequently the NT4.0 server ...
> I was running PINGs across the real IP adresses and across the tunnel IPs.
> ON start of test real IP ping was about 40-50ms and stayed there all the
> time. The tunnel started with 50-60ms and slowly goes up until 150-220ms
> and then single pings are dropping out >3000ms. That gets more and more
> often, until the tunnel crashes.
> And, if the tunnel had crashed, unloading CIPE from the services results in
> a sudden death of the NT4.0 box ...
> I'll try to run a sniffer on both sides of the tunnel to get a clearer
> picture of packer delays.
> Could it still be a MTU/fragmentation problem?
> I'm running 1450 as MTU on the DSL-line and 1330 on the CIPE interfaces ...
> best regards,
> >I've got VNC running here, I'll try it out. Did that happen with pre14 ?
> >I still need to apply the changes to the Win2k install registry settings
> >provided by Marco Palumbi <marco.palumbi,AT,bigfoot,DOT,com>. Those should
> > with the startup failures.
> >On Saturday 26 October 2002 08:00 am, Jürgen K. Zick wrote:
> > > Hi Damion, hi all,
> > >
> > > since I run the most recent NT version of CIPE (2.0-pre15) I have a
> > > strange crash problem:
> > >
> > > CIPE runs very stable on the NT4.0 server and the associated W2K-remote
> > > workplaces. Nearly perfect, after I made a MTU fine tuning. I made huge
> > > fine transfers, tested heavy TCP and UDP traffic, all worked fine ...
> > > BUT, there is one occasion, when the tunnel goes down. If I do a remote
> > > control with VNC over the CIPE tunnel to a machine on the "hardwired"
> > > ethernet interface behind the server the tunnel will break down within
> > > minutes, and really only the tunnel, nothing more. No error message at
> > > all, it just stops to work. Then, when I try to stop the CIPE service
> > > on the server in that state, the NT4.0 server crashes silently without
> > > any blue screen ...
> > > Most irritating, when I use VNC over the tunnel to control the server
> > > itself, it is fine. When the VNC packets have to be passed further on
> > > by the server to the network, then it crashes the link ...
> > >
> > > By the way, the CIPE service, also with this distribution, still does
> > > not start more than 50% automatically after a reboot ...
> > >
> > > So, any ideas what could be improved to avoid the crashes?
> > >
> > > best regards, Jürgen
> >Message sent by the cipe-l,AT,inka,DOT,de mailing list.
> >Unsubscribe: mail majordomo,AT,inka,DOT,de, "unsubscribe cipe-l" in body
> >Other commands available with "help" in body to the same address.
> >CIPE info and list archive:
> I FFFFF A Jürgen Zick, Dipl.-Berufspäd. Dipl.-Ing. VDI
> I F A A FB15 - Institut für Arbeitwissenschaft (IfA)
> I F A A Tel.: (0561) 804-4497 Fax: -4162
> I FFF AAAAA Universität Kassel
> I F A A Heinrich-Plett-Str. 40 - D-34132 Kassel
> I F A A *** SOKRATES - ERASMUS - EXCHANGE KOORDINATOR des IfA ***
> I F A A *** EDV-Netzmanagement des IfA ***
> alt. Email: jzick,AT,uni-kassel,DOT,de