How to reduce a very high latency number

[Trask]

Ok, I've decided to not wait before addressing the "tuning" doc
(http://nagios.sourceforge.net/docs/2_0/tuning.html).  Also, thanks
for the link Jim, there is some good information there.

(Doc)   1. Use aggregated status updates.
(me) Doing this.

(Doc)   2. Use a ramdisk for holding status data..
(me) Not doing this -- the doc says if "you're not using aggregated
status updates", so I skipped it.  Does anyone have any experience
with how helpful this would be?

(Doc)   3. Check service latencies to determine best value for maximum
concurrent checks.
(me) Currently set at 500 -- I've set it to 0 (no limit) and I have
done all the calculations and set it to 15% above the number of
concurrent checks I should need based on the timing.  Doesn't ever
seem to have any effect on latency.

(Doc)   4. Use passive checks when possible.
(me) About a third of our checks are passive (nrpe, that's considered
passive, right?).

(Doc)   5. Avoid using interpreted plugins.
(me) I use built-in plugins whenever possible, but otherwise it is a
Perl script 90% of the time.  Some of them are pretty complex, but...
(see #6)

(Doc)   6. Use the embedded Perl interpreter.
(me) I am doing this and using the "perlcache" option.  In fact, I
haven't even tried our setup without using the embedded perl
interpretter, so I might end up trying that out.

(Doc)  7. Optimize host check commands.
(me)  Did this -- currently using check_fping with "-n 1" and retrys
at 8.  I don't actually have any numbers before I did this, so I don't
know how much help it has been, but when I first read about it, I was
hoping for a lot more gain than I got...

(Doc)   8. Don't schedule regular host checks.
(me) We don't.

(Doc)   9. Don't use agressive host checking.
(me) We don't.

(Doc)  10. Increase external command check interval.
(me)  This is set to -1 right now on all of our hosts, but this is
probably only sensible for our central server (nag4).  I will
experiment with changing this to 10 seconds on our other servers and
see if that helps.

(Doc)  11. Optimize hardware for maximum performance.
(me)  P4 2.4GHz, 512MB RAM, 7200rpm IDE drive w/ DMA enabled.  Is
there a good way to see the current IO activity of the disk to see if
it is being hammered?  "top" shows the CPU and memory at very
reasonable levels.


Anyone have other tips?  Do my responses to these tuning
recommendations seem to make sense?


On 5/17/06, Jim Perrin <jperrin at gmail.com> wrote:
> > Ok, that is all I can think of off the top of my head.  I have
> > reviewed the performance tuning tuning doc (from here:
> > http://nagios.sourceforge.net/docs/2_0/tuning.html), but I am open to
> > trying things again / in a different way.  I can list off what I've
> > done in response to that doc on a point-by-point basis if anyone is
> > interested.
> >
> > Thanks for any help -- this latency issue is the last big hurdle
> > before getting this thing going.
>
> At the risk of sounding fanboy-ish, there's a metric assload of decent
> nagios information here -> http://altinity.blogs.com/  including
> several good patches that help to tweak nagios in various ways. Some
> of the last couple entries have involved tuning that may help.
>
>
> --
> This message has been double ROT13 encoded for security. Anyone other
> than the intended recipient attempting to decode this message will be
> in violation of the DMCA
>


-------------------------------------------------------
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