starting a service on use

[Lennart Sorensen]

On Fri, Oct 08, 2010 at 10:05:17AM -0400, Christopher Browne wrote:
> That's a pretty good answer.
> 
> If someone really doesn't want the service automatically started, they
> might look into "port knocking" as an approach...

True.  After all the port knocking won't be a failed connection to the
real service, it is a seperate step.

> Basically, you have a service (possibly xinetd-based, or watching logs
> for port denials) which watches for someone knocking at some sequence
> of ports.  If it sees this, it starts up the "real" service.
> 
> But that requires a funky protocol addition for the client that was
> going to try to connect; it needs to be modified to "knock on the
> door" first.  Which is quite likely to be an unacceptable imposition.
> 
> http://en.wikipedia.org/wiki/Port_knocking

Of course if you are going for simple and you are the only user that
needs the service, how about:

ssh to box, start service, connect to service.  Seems simple.  If you
need it often enough that that isn't convinient, then you should probably
just leave it running.

-- 
Len Sorensen
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists