TCP-forwarding with netcat - weird failures ...

[Patrick M. Hausen]

Hi all,

in a particular customer network we have a world wide VPN with partially
overlapping addresses, renumbering impossible due to political reasons,
and all the fun you can have in the „enterprise“ environment.
No IPv6 either, newfangled nonsense … :-/

So to access a certain set of services we installed a VMware virtual
machine running FreeBSD at the VPN’s central hub and users in
subsidiary offices use the IP address and certain ports on this machine
which then hands off to the target service that is not reachable from
the subsidiary.

Setup is dead simple, just one example:

/etc/services:
oediv-3243	3243/tcp

/etc/inetd.conf:
oediv-3243	stream	tcp	nowait	nobody	/usr/bin/nc	nc 172.20.1.166 3243

The machine has been in service for 10+ years and runs FreeBSD 6.4.

So what’s the problem? The VMware environment that hosts this machine
is about to be retired. So I installed a fresh VM with FreeBSD 11.2 plus
current VMware-tools and copied the setup.
Then we shutdown the old machine and booted the new one with identical
IP address.

Needless to say: doesn’t work. And no, it’s not the obvious ARP caches.
Connections can be established but then abort spontaneously without
an observable pattern or reason.

We already found that more modern netcat/nc needs „-N“ to close the
connections on EOF, but besides …

Does anyone know what might have changed that could cause connection
problems?

Kind regards,
Patrick
-- 
punkt.de GmbH			Internet - Dienstleistungen - Beratung
Kaiserallee 13a			Tel.: 0721 9109-0 Fax: -100
76133 Karlsruhe			info at punkt.de	http://punkt.de
AG Mannheim 108285		Gf: Juergen Egeling