Kaseya Community

ESX CPU Monitoring

  • Hello all,


    I haven't found a decent way to monitor CPU's from ESX hosts.
    We have ESX servers with 16 cores (32 hyperthreaded) and it seems i can monitor utilization per core, so i'd have to add 32 monitos to a template serperately per core.


    Isn't there a way to add all 32 in one monitor, so you have let's say, 32 lines on one chart, and if one of 'm is higher then 95% for longer then 5 minutes, we get an alarm ?


    Also, are there any more (community driven?) lua scripts available somewhere?


    Thanks in advance!



    additions
    [edited by: ACA-IT at 6:08 AM (GMT -7) on Aug 7, 2013]
  • There's a limitation, due to performance issues, with having that many channels of data per monitor. Currently, you can have a maximum of 8 channels per monitor, meaning you could have 8 cores per monitor. What IS currently possible though, is creating a Lua script that would gather the data from all cores and do what you are asking for. It would not right now be able to save the statistics for each core though, but would be able to set an alarm the way you want it to.

    Sorry, but currently we don't have a community driven Lua or device template library but it's on the roadmap.

  • Thanks for the answer!

    How can i add 8 cores to one monitor? i can't seem to find the option to add additional monitors to a monitor.

    Also, is there a way to remove the discovery group ? I have disabled autodiscover as i don't want to use it, and removed all the discovered devices, but the group stays there.



    Spelling errors
    [edited by: ACA-IT at 5:14 AM (GMT -7) on Aug 9, 2013]
  • Adding 8 in one monitor would be through a Lua script, but there isn't a really good way of doing that now. We have not exposed the VMware performance library to Lua yet so the only way would be through SNMP at the moment, if that's available.

    The discovery group is created when Discovery is run for the first time. There's no current way to remove the group, once it's created. That would have to be considered as a feature request in that case.