RE: [bug-inetutils] [patch] inetd: set environment variables for remote host, ip and local port

[Alain Magloire]

> From: address@hidden [mailto:bug-inetutils-
> address@hidden On Behalf Of Alfred M. Szmidt
> Sent: Thursday, July 27, 2006 4:22 AM
> To: Dirk Jagdmann
> Cc: address@hidden
> Subject: Re: [bug-inetutils] [patch] inetd: set environment variables for
> remote host, ip and local port
> 
>    > Wouldn't it be better to provide this information in log (if it
>    > isn't provided already)?  And then just use grep to get that
>    > data? I'm a bit sceptical to exporting envars like this.
> 
>    Reading from syslog will lead to concurrency problems if multiple
>    server processes are spawned from inetd simultaneously.
> 
> Each process could provide its own log, on a heavily trafficed machine
> that should be done anyway.
> 
> Frankly, I atleast am not for or against adding this (I can't really
> see the benefit of such a feature, but I can't see why adding it would
> be bad).  Maybe Sergey or Alain can pitch in.
> 

I was trying to think are there any security problems with this, but could
not find anything obvious.

So we could add this and turn it on when you start inetd with particular
options.  Reverse lookup can be extremely expensive, so agreed it should be
configurable.

This is an interesting problem in general; how to set particular env
variables.  Xinetd solves this by having a new more flexible xinetd.conf but
the file is not backward compatible.  Maybe we should look at providing a
compatible xinetd.conf parser for inetd.


IMHO, if we can not find any security flaws to a feature, I do not see any
reason to refuse it.

P.S.: I like Rich ideas to follow the D. Bernstein style.

> Cheers.
> 
> 
> _______________________________________________
> Bug-inetutils mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/bug-inetutils