Ad Widget

Collapse

CPU and Memory hotspots for server environment

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • BDavisJD
    Junior Member
    • Aug 2016
    • 2

    #1

    CPU and Memory hotspots for server environment

    I was curious if anyone has a graph that will display a small square for each server in the environment. Ideally i would like 2 sections, one for cpu and one for memory. Low CPU usage is Green and high CPU is Red, same for memory. Each squad may be 10 pixels by 10 pixels

    I had tried with screens some, but it appears that I have to implement every server manually and I would need this to be dynamic as we add more servers to the environment.

    I'm still learning Zabbix though...
  • BDavisJD
    Junior Member
    • Aug 2016
    • 2

    #2
    I guess i'm looking for a heatmap of the environment as a whole

    Comment

    • kloczek
      Senior Member
      • Jun 2006
      • 1771

      #3
      There is no OOTB solution for what exactly your need because first you need to define what is "high" and what is "low".
      Sometimes as well organizing alarming layer on top of memory usage on OS layer is pointless as applications are starting allocating almost 100% avalaible memory to maintain memory management on application layer.

      In such conditions usually OS memory usage alarm is defined as less than 98% memory usage or more than let's say more than 1GB free memory. Such alarm can be used to confirm that application just died or someone overprovisioned OS memory or undeprovisioned application memory usage.
      As you see the alarm may have in some exact conditions completelly opposite logic definition.

      In other words adding alarming layer will be done on top of completly different set(s) of metrics.

      You may have as well more than two typical memory consumers because on some systems kernel memory usage may be not so statius like on typical Linux system.
      For example on Solaris you probably want to know not only about kernel memory usage but probably ZFS ARC (Adaptiive Recalim cache) memory usage.

      The same is with CPU usage. I know places where "high CPU usage" is defined on top of number of CPU context switches/s (as more than 2k*number of CPU cores).

      On presentation layer you have triggers overview with stats of the triggers and you can organize exact page to display in two columns presented as coloured squares.
      http://uk.linkedin.com/pub/tomasz-k%...zko/6/940/430/
      https://kloczek.wordpress.com/
      zapish - Zabbix API SHell binding https://github.com/kloczek/zapish
      My zabbix templates https://github.com/kloczek/zabbix-templates

      Comment

      Working...