latency and parallelize check

[Marc Powell]

> -----Original Message-----
> From: nagios-users-admin at lists.sourceforge.net [mailto:nagios-users-
> admin at lists.sourceforge.net] On Behalf Of Trask
> Sent: Thursday, April 27, 2006 12:36 PM
> To: nagios-users at lists.sourceforge.net
> Subject: Re: [Nagios-users] latency and parallelize check
> 
> On 4/27/06, Sandeep Narasimha Murthy <sandeep-n-murthy at telecom.pt>
wrote:
> > <snip>
> >
> > What about latency, any suggestions on how I can better it !?
> >
> 
> I'm very interesting in this, too.  Is there any information as to
> what each of the "performance" stats really mean?  I'll preface by
> saying I have not gotten around to setting up any external performance
> monitoring...
> 
> Here are some more specific questions to get things started:
> 
> The 4 tables on the left side of the "Performance Info" page have 2
> columns -- "Time Frame" and "Checks Completed".
> 1) Does the "Time Frame" start from when a check is scheduled or from
> what it actually starts?

>From when you loaded the page. They show the number of checks that have
completed in the last 1, 5, 15 and 60 minutes.

> 2) If nagios has been running > 24 hours and I still have less than
> 100% "checks completed" for the "since program start" row, what does
> this mean?  Do I have checks that are never ruturning (doesn't seem
> possible given the built-in timeouts).  Are there checks that are
> getting used that are getting factored in?

It means that you have some percentage of checks that aren't being
executed. If these are service checks this could be a configuration
issue. Sort your Service Detail by Last Check and see if you have ones
that haven't been updated.
 
> Metrics:
> 3) Does "check latency" mean how long after a check is scheduled that
> nagios gets the result?  Or after it has completed?  Or after the
> check has started?  (For reference:
> http://nagios.sourceforge.net/docs/2_0/checkscheduling.html )

It's the time between when Nagios originally scheduled it to run and
when it actually did run. If you see latencies > 10 seconds or so it
probably means that you're not allowing nagios to run enough concurrent
checks.

> 4) Cleary, we want lower execution times and low latencys -- what are
> people who are satisfied with the performance of their setup seeing
> for these numbers?

I'm running on some really old hardware for most of my collectors that I
am in the process of replacing so the only box I'm really satisfied with
is performing 498 checks with an average Execution time of 3.3 seconds
and latency of 0.161 sec. The others are pretty high right now but
complete ~1500 checks each in my scheduled 5 minutes so I haven't been
too concerned about latency or execution time.

> 
> General:
> 5) Is there any way built in to nagios to get performance data on each
> check for ALL hosts? This is in the spirit of "checking the check" to
> see if it is performing reasonably well as well as serving as a basis
> to compare the performance of different checks against one another (o
> r comparing the difference between a compiled plugin's performance
> verus a non-compiled one).

Not currently that I am aware of. Maybe this is something that would be
possible via the Event Broker.

> 6) Any suggestions for 3rd party addons, plugins, etc that could help
> shed some light on to these performance questions?

No suggestions here. I've never looked myself...

--
Marc


-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid0709&bid&3057&dat1642
_______________________________________________
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