2 Replies Latest reply: Jan 11, 2013 2:57 PM by tolga.asik RSS

VM disk latency - alert

tolga.asik

I configured this alert to get alert e-mails after 30 ms latency encountered. But VMAN sends me an alert when under 30 ms latency encountered.

 

Search Query : vm.powerstate:poweredOn AND (vm.readLatency.latest:[30 TO *] OR vm.writeLatency.latest:[30 TO *])

VMdisklatency-alert.jpg

Notification value on this alert :

VMdisklatency-NotificationValue.jpg

How can I remediate this ? I don't want to get this alert e-mail under 30 ms.

 

Could you help me please?

 
  • Re: VM disk latency - alert
    mbartak

    Hi, the alert as is defined is correct. It returns only running VMs with read or write latency > 30ms. You can validate that by performing same search you have defined in the alert and move mouse the the hits in search results.

     

    In the notification value you are showing summary of write and read latency divided by 2 and this cause the numbers under 30. E.g. write is 40 and read is 10. writeLatency is enough to trigger the alert but notification value will show 20.

     

    I propose to change notification value.

    • If you wanna see just hight of latencies (e.g. write: 50ms):

    if(/virtualMachine/readLatencyUtilization/latest/text() > /virtualMachine/writeLatencyUtilization/latest/text())

    then (concat('read: ',/virtualMachine/readLatencyUtilization/latest/text(),'ms'))

    else (concat('write: ',/virtualMachine/writeLatencyUtilization/latest/text(),'ms'))

    • If you wanna see both values (e.g. : read: 5ms; write: 50ms)

    concat('read: ',/virtualMachine/readLatencyUtilization/latest/text(),'ms; write: ',/virtualMachine/writeLatencyUtilization/latest/text(),'ms')

     

    In both cases you need to change Unists to 'none'