Nagios2 on FreeBSD 5.3

[Stanley Hopcroft]

Dear Sir,

I am writing to thank you for your letter and say,

On Wed, Feb 09, 2005 at 08:09:44PM -0800, nagios-users-request at lists.sourceforge.net wrote:

> Message: 31
> Date: Thu, 10 Feb 2005 01:33:23 +0000
> From: Haze <nagios at virtualhaze.org>
> To:  nagios-users at lists.sourceforge.net
> Subject: [Nagios-users] Nagios2 on FreeBSD 5.3
> 
> Hello all,
>   I am looking at moving our Nagios 1.2 clusters to Nagios 2 and I am 
> asking if anyone has had any experience as to how well it works on 
> FreeBSD 5.3.
>

Unfortunately no (yet, maybe) but since the KSE feature is in '5, this 
is certainly worth trying.

Thread support in '4 seems pretty average from this dumb users 
persepctive.

It's only after going to a threaded Perl (5.8.5, built from ports), that 
I have witnessed the massive leak rate in embedded Perl Nagios that 
others have complained about.
 
> In brief we have 1 logger/cgi server and 4 remote servers with 60, 80, 
> 500 and 2000 hosts. Total services across all the remote servers is 
> ~10000. The top heavy service to host ratio is due to 90% of the devices 
> being routers, where we check for all interfaces that at in an ifAdminUp 
> status (nagios conf population is done automatically) and all the env 
> stuff (about 5 on most routers). All done via Perl Net::SNMP and ofc 
> embedded Perl.

Glad to hear that there are at least a few sites that make heavy use of 
what I consider a valuable feature (and wouldn't start Nagios without 
it).

<even more off-topic>
If you don't mind me asking, do you have a Net::SNMP process running as 
a monolithic (ie polling all your routers) poller in non-blocking mode 
and submitting passive service check results to Nagios ?

Maybe not since it would be hard to synchronise the configs unless your 
poller was re-reading the Nag configs with perhaps Nagios::Config.
</off topic>

You should be aware that the embedded Perl support in Nagios 2 _may_ 
have taken a step backward from that in 1.x.

At least one large embedded Perl Nagios (ePN) site has reported a much 
greater leak rate (presumably with the same Perl and platform) after 
going to 2.x. Sites of your magnitude, as it stands, have no option but 
to monitor the size of the Nagios process (with check_procs/check_vsz) 
and restart Nagios if the memory usage is over threshold. Either that or 
disable Perl caching and take the performance hit.

Testing is underway now to try and determine, by reversing the changes 
in ePN, whether this is caused by the ePN changes or something else in 
Nagios.

If you choose, please keep the list (or me) informed of how you go. If 
the testing shows that there is a simple fix (by backing out ePN 
changes), you are welcome to try them.

> 
>  From the docs there looks like there is an issue with pthreads with 
> large setups, is this still so?
>

I think I have witnessed the occasional stagnant service check that I 
think is caused by this (FreeBSD 4.10-RELEASE). This too on a modest 
setup (200 hosts/300 active/50 passive).
 
> Any other advice or pitfalls to avoid would be gratefully accepted.
> 
> Thanks
> 
> -------------------------------
> Haze
> Synetrix PLC
> -------------------------------
>

Yours sincerely. 

-- 
Stanley Hopcroft

IP Australia
Ph: (02) 6283 3189  Fax: (02) 6281 1353
PO Box 200 Woden  ACT 2606
http://www.ipaustralia.gov.au
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: disclaimer.txt
URL: <https://www.monitoring-lists.org/archive/users/attachments/20050215/1ed6a2f2/attachment.txt>