[Nagiosplug-help] How to block nsca when central nagios server is down?

[Ralph.Grothe at itdz-berlin.de]

Hello Nagios Users,

I have a nagios 2.5 central server running on a somewhat
overloaded old AIX PowerPC.

The haevy load increase when nagios simultaniously schedules
several checks 
(especially those Perl checks such as check_squid; no I haven't
built nagios with emb_perl)
was the reason why I was asked by colleagues who were impaired to
at least temporarily shut down the nagios server
(I am desperately waiting until I get some piece of HW for a
dedicated nagios server).

I know this sort of "part time monitoring" must sound quite
bizarre, but I am extremly resource restricted.

Now, as long as the central nagios server is down passive checks
by send_nsca from distributed nagios servers,
that I didn't care to shutdown as well, keep connecting to the
nsca port and having inetd spawn ever more new processes
until the user nagios reaches the max user procs kernel limit.

Of course I can become root and kill those idle nsca children.
But wouldn't it be better to block any requests to the nsca
socket while the nagios server is down?

I could write a stop/start wrapper script for the nagios server
that would reconfigure inetd to either ex-
or include the nsca server and send it a SIGHUP afterwards.
But as I run the nagios server under uid nagios I would also need
to set up a sudo rule for nagios
to reconfigure inetd.

Is there a better way for this kind of weird scenario,
something that doesn't require elevated privileges?


Also would I need some advice how to tune the nagios server's
performance
so that its check wouldn't blow up the load.
In the online docs I read that one could set
max_concurrent_checks from 0
to some low enough fixed value that on the other hand still would
accommodate enough room for all the checks
to be performed in their normal check intervals.



Rgds.