No announcement yet.

Expected Interface Performance

  • Filter
  • Time
  • Show
Clear All
new posts

  • Expected Interface Performance

    Hello folks! We have just setup Oreon as a trial here to see if we like the interface, and so far it looks great. I do have a question about performance of the interface... can anyone tell me what I should expect? An example is the service problems page. 5 problems are listed on the page... We have right now 60 hosts with 1100 services.

    Result is 15 seconds for the page to render. It takes 2 seconds on the native nagios cgi interface. I realize that the cgi is natively compiled machine code vs runtime php.... but 15 seconds is a fairly long time to wait

    Hardware? Sun T2000 8 core Sparc .... A killer box for multithreading, but terrible on single threaded performance. I assume my problem is just that, the interface is very single threaded in its core.

    Anyway, I just wanted to know what everyone else is seeing in performance ...comparing the native nagios service problems page against the oreon page.


  • #2
    Hi, from what I see on my station there is little difference between the loading time of the nagios cgi vs the oreon php page. i.e. oreon services page with 200 services displayed loads in approx 2 seconds. (desktop with 2.6GHz Celeron)

    have you tried checking the load average of the sun box ?


    • #3
      I have 500+ hosts with 600+ services and my load has never exceeded 2 seconds. Granted I'm running a Dual Proc Xeon 2.60, 2GB memory, 10,000RPM SCSI drive with nothing else running but Cacti.
      Does seem a bit excessive to be seeing 15 seconds. Especially when your Nagios interface is much quicker.


      • #4
        you could try increasing memory limit to 16 or 32 Mo in php.ini and restart apache

        I agree the page should not take this much time to load

        as for the load average, can you confirm the values of the command 'uptime' for example, because why "seconds" ?


        • #5
          php Memory limit was set to 1 GB, so no worries there. My load average? 1.5

          I have narrowed it down to the code that processes the status.log, which is about 1.5 MB in size. I noticed the fast status option, which has zero documentation I can find... however I figured out what it was looking for, and wrote a nifty little C++ program to take the status.log and output the status.log_light file in the format it wants,\. That changed my render time from 15 seconds to 7 seconds. Big improvement, but still 5 seconds slower than the nagios native interface.

          Anyway, thanks for the help everyone, any other ideas are appreciated!