[l2tpns-devel] Some patches: Makefile (again), ppp_terminate_from, and daemonize.

[Juergen K. <j.k...@eu...>]

Hi David,

find three patches as attachments.

1. Makefile.patch
    My last patch there saved existing config files from being overwritten.
    But I did not correct the names for config files when installing for
    the first time.
    My reasoning is: provide new users with a working config, even when
    the default values in it may not be correct for them.  Warn them
    during the install.
    I find this behaviour better then to install files with other
    names (l2tpns.users etc.) which are not used at all - and resulting
    in the "missing config file" error when you start l2tpns as a novice
    without configuring at all.

2. Terminate_From.patch
    The manual was a bit incomplete about the "bind_address" parameter.
    I realized the hard way that I could not set the gateway address
    provided for the ppp connections separatly from the bind_address,
    which is used for the l2tp tunnel termination.
    After applying the patch, it runs as follows:
    If "terminate_ppp_from" is set, that address is used for the
    ppp gateway address.
    If it is not set and bind_address is set, bind_Address will be
    taken for the ppp gateway address (this is to stay compatible).
    Final ppp gateway address (as before) is the hardcoded 1.1.1.1.

    Documentation update is included.

3. Daemonize.patch
    The option "-d" for daemonizing is listed when calling l2tpns with
    wrong parameters, but it was not recognized yet (missing in the
    getopts).
    I cleaned up the daemonizing a bit while realizing that.

Beside these patches, I took your "generate_load" test program as a base
for setting up a test environment.  After quite some fiddling I got it
to talk to a cisco lns.  I think I will be able to get "hard" numbers
for the cisco lns in direct comparison to l2tpns during next week.

Maybe generate_load and bounce should be brought into cvs too - a
contrib directory or something similar.  Generate_load
may be quite far away from being usable as a lac, but it has
the potential to be transformed into such a beastie.

I have l2tpns debianized in the meantime.  If there is interest to
include this directly into cvs, I can provide a patch for that.

	Regards,
		Juergen.