Patch from Ralf Paffrath <paffrath@dfn.de>.

Patch from Ralf Paffrath <paffrath@dfn.de>.

RADSECPROXY-37.

Related to RADSECPROXY-37.

TODO: Add documentation.

Related to RADSECPROXY-37.

TODO: Add documentation.

Closes RADSECPROXY-40.

This is the old number.  We used 1 minute during testing.

Closes RADSECPROXY-36.

This goes in branch dynconf2.

This reverts commit dbcc9977.

Dynamic servers has clearly never been run on UDP servers.
We should probably do something less evil than crashing here.
Closes RADSECPROXY-26.

clientwr() should treat the dynamic lookup as a failure and will not
be any happier to know that the exact error was that it didn't resolv.
The script can do whatever logging is wanted.

That said, this commit also makes the scripts exit with 10 in order to
signal failure.

Also, sleep less than 15 minutes (900s), mainly for testing.  This
number will change.

Also, die hard and explicitly if freeing an already freed config in
freeclsrvconf().

This is part of fixing RADSECPROXY-33.

Using /bin/bash isn't portable.

`-e' to /bin/echo isn't portable.  The BSD's lacks it for instance.

Spotted by Paul Dekkers.  Closes RADSECPROXY-39.

This closes RADSECPROXY-36.

Note that `-i' on the command line overrides this config option.

This closes RADSECPROXY-32.

This closes RADSECPROXY-31.

Much like what 1c05812c did for radsecproxy.conf-example.

We change from "mysecret" to "radsec" as per
draft-ietf-radext-radsec-12.txt section 2.3 (4).

In the error case in confserver_cb() where compileserverconfig() we
used to go to errexit and have resonf, passed as an argument through
ARG, freed.  Other parts are still using that conf.