BPI is maxing out CPU

[whjsmith]

Anytime we attempt to view the BPI, create a new BPI group or edit existing members the httpd service CPU usage goes to 60-95% and the web pages take 13+ seconds to load. We currently have 2 cpus on the server. We also have 10,000+ service checks on the server (checks are distributed). It has to be the number of service checks that is causing the issue. We setup the identical BPI on our test system that only has 300 checks and it is quick. We copy that same BPI over to our production system and it grinds to halt. Where does the BPI query to get the results of the service checks that are associated to the BPI? Is it from the Nagios database?

Thanks for your help,
Jason

[scottwilkerson]

This is well beyond the amount of services that is recommended to be checked with this hardware. Anything over 3000 checks we recommend have > 4 CPUs

http://assets.nagios.com/downloads/nagi ... ements.pdf

[whjsmith]

Ok I will up the resources and see if that helps.

Jason

[whjsmith]

Oh wait... sorry. That document you linked to is for a server processing active checks. We distribute the checks to 6 other servers. That server that is displaying the BPI does not run any checks. It just processes the results. Back to my original questions.. does the BPI query the Nagios database to get the results of the service checks?

[mguthrie]

Which version of Nagios BPI are you using? The 1.3x version will use a LOT of CPU on a large system for each check, whereas the 2.x version caches data so it's much more efficient.

[whjsmith]

We are using BPI 2.0 on Nagios XI 2012R1.1

[mguthrie]

Can you make sure the cache file has write permissions on it? If for some reason it wasn't writable then the checks would definitely be a big CPU grabber.

Code: Select all

chown apache.nagios /usr/local/nagiosxi/var/components/bpi.xml
chmod 775 /usr/local/nagiosxi/var/components/bpi.xml

[whjsmith]

The bpi.xml file had 644 permissions and was empty. I have changed it to 775 permissions 4 hours ago but the file is still empty. How do I get the BPI to start caching the data? We also added 2 cpus and 4 gigs of RAM to the server. It helped maybe shave off a couple seconds but the BPI still takes 10+ seconds to load.

[scottwilkerson]

Can we take a look at who the user/group are for the file

Code: Select all

ls -l /usr/local/nagiosxi/var/components/bpi.xml

[mguthrie]

Can you also verify that particular location is specified in the BPI Options? Check the Admin->Manage Components page, and click the edit settings icon for the BPI listing. BPI Supports a custom location for that file, so it would be good just to verify that the location matches.