Hi, I'm looking for the same kind of tips : I've a P4 2.8GHz, dual-proc/hyperthreading, 512 Mo RAM, 2 Go swap, and near than 4000 services on 300 hosts. (embedded-perl snmp checks + rrd bases access in most of the case) and i have trouble with performances : - average check latency is 20min - swap is almost full, - interface web is slow.. Well, after tuning, CPU is now ok (thanks to max_concurrent_check=350) use, but latency is always too important ; same thing for swap usage problem. Any idea to improve things will be apreciated.. Thanks. Regards, -- Gilles POIRET <table width=100%> <tr valign=top> <td> <td>"Platt, Nicholas" <Nick.Platt@myflorida.com> Sent by: nagios-users-admin@lists.sourceforge.net 21/04/2004 16:34 <td> To: "'nagios-users@lists.sourceforge.net'" <nagios-users@lists.sourceforge.net> cc: "Tedder, Brandon" <btedder@bearingpoint.net>, "Radcliffe, David" <david.radcliffe@myflorida.com>, "Chelette, Robert" <robert.chelette@myflorida.com> Subject: [Nagios-users] Memory size question on a single nagios process</table> Hello everyone: I've been working on tuning our production server running Nagios which monitors currently 1500 services and will eventually monitor 5000. I was able to lower the CPU spikes and keep Linux from swapping by lowering the max_concurrent_check=325 and command_check_interval=2 based on the formula depicted in the documentation. We are currently monitoring the latest check time to the earliest check time and we are find that it is an average of 8-10 minutes which currently is not acceptable. Any recommendations on minimizing that time would be appreciated. Linux System: P4 2.8 Ghz with 768 RAM My second question has to do with why individual Nagios check processes take up 3172KB of memory. That's a lot in my opinion especially if you multiple that times the number of checks. Is there an explanation? Can it be lowered? See attachment. Thanks