Re: Cfengine ate my mail

[Mark Burgess]

I think its a bit much to blame this on cfengine. Relying on a network
service is always fragile.

However, I do think this is a good argument to put exception handling
into version 3.

M

On Tue, 2005-08-16 at 17:26 +0100, Andrew Stribblehill wrote:
> We run Cfengine across all of our servers and workstations in Durham.
> One of the things it does is control the sendmail.cf file and HUP
> sendmail if it changes.
> 
> To determine whether or not a host is a mail exchanger, we have:
> 
> groups:
>   mxes = ( address@hidden )
> 
> However, we lost access to NIS (the core router blew up) so our MXes
> forgot that's what they were and reconfigured themselves, presumably
> caching the negative response in nscd. Net result: we bounced a load
> of legitemate mail.
> 
> I believe that Cfengine could do better than this, by checking the
> return code of setnetgrent() and (conditionally, perhaps) aborting if
> it can't find the netgroup.
> 
> What do other people think?
>