NEED HELP configuring NSCA with INETD

[Paul L. Allen]

Michael Tucker writes: 

> As a side thought... what are the costs and benefits of running nsca
> under inetd as opposed to as a standalone daemon?

Standalone daemon will read its config once at startup; under inetd
it reads the config for each check.  Standalone daemon has the overhead
of loading itself into memory once at startup, and the overhead of a
process fork for each check if it's a forking daemon, the overhead
of being swapped back into memory if it was swapped out; under inetd it
has the overhead of loading itself into memory for each check (and
inetd might be swapped out if that's all you use it for).  Standalone
daemon should accept the connection and return the results slightly
more quickly.all other things being equal.  Standalone daemon consumes
a certain minimum amount of memory/swap all the time, with intermittent
spikes if it's a forking daemon when it receives results; under inetd
there is no constant memory/swap use from ncsa (but there is the constant
minimum of inetd) and there are the intermittent spikes when it receives
results.  Standalone daemon may have memory leaks and/or no mechanism
for ensuring it is always present (I haven't run it or looked at the code
to see if there is some sort of phoenix mechanism); inetd probably has had
all memory leaks found by now and it designed to ensure it is always
running. 

Which one is better?  Try it and see, because it will depend on whether
you're using inetd just for ncsa or for other things too; whether the
machine is using swap space and all sorts of other things.  There's
probably not much in it in terms of resources consumed because the
footprint isn't that big anyway.  If you receive a lot of results then the
(theoretically) slightly slower response under inetd might become an
issue.  If you find the standalone daemon turns out to have a memory leak
or dies every so often (I've not seen anything to suggest that it has
either of those problems but just about every daemon has had problems
of those kinds at one time or another) and you have to manually start it
again then inetd might be better from the viewpoint of reliability. 

BTW, most people seem to have abandoned inetd in favour of xinetd which
is supposed to be an improvement on inetd. 

Oh, and the last time I played with NRPE (same author) I couldn't get
it to run standalone so had to run it under xinetd whether I wanted to
or not.  I probably missed something obvious, but I was under pressure
to get it working quickly so I went along the path of least resistance.
Sometimes the question is not which way is better but which way can you
get it working before the PHB blows his top over how long it's taking... 

-- 
Paul Allen
Softflare Support 



-------------------------------------------------------
This SF.net email is sponsored by: Perforce Software.
Perforce is the Fast Software Configuration Management System offering
advanced branching capabilities and atomic changes on 50+ platforms.
Free Eval! http://www.perforce.com/perforce/loadprog.html
_______________________________________________
Nagios-users mailing list
Nagios-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. 
::: Messages without supporting info will risk being sent to /dev/null