Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Table of Contents

Node Health Report

The Node Health report display health-related attributes for all selected nodes for a given period. Attributes displayed are: Status, Device, Availability, Interface Availability, %CPU, 95th% CPU, Max %CPU, CPU Exc., %Mem Free, 95th% Mem Used, Max %Mem Used, %Mem Util, %IO/VIR Mem Free, 95th% IO Mem Used, Max %IO Mem Used, %IO/VIR Mem Util. As of version 3.1.4 when this report is exported to XLSX and CSV formats the following columns of information are also displayed: Group, %IO Mem Free.

...

If the node has multiple CPUs then the utilisation measure is averaged over all CPUs. 


CPU Exceptions
The count of times the CPU utilisation exceeded the "CPU Exception Threshold" of 20%. If the node has multiple CPUs then this is the sum of the exception counts of all CPUs.

...

  • Low free main memory less than 25


opReports 3.5.1 and newer include improvements to memory related fields

/path/to/omk/conf/opCommon.json has a new setting withdefault being /opreports/on_invalid_hrcachemem_use_only_hrmem=0.

  • With /path/to/omk/conf/opCommon.json set at opreports/opreports/on_invalid_hrcachemem_use_only_hrmem=1,
    opReports attempts to detect situations where invalid hrCacheMemUsed and hrCacheMemSize
    values are being reported that would cause the memory related fields in the Node Health Report to return as negative values.
    • In such a case (detect memory related fields in the Node Health Report would negative values),
      hrCacheMemUsed and hrCacheMemSize will not be used in the calculation of memory related fields
      and a suitable entry to this effect will be logged in opReports.log.
    • This issue has been detected in docker instances where hrCacheMemUsed and hrCacheMemSize values
      were that of the docker host and not that of the docker instance itself.
  • With /path/to/omk/conf/opCommon.json set at opreports/opreports/on_invalid_hrcachemem_use_only_hrmem=0:
    • In such a case (detect memory related fields in the Node Health Report would negative values),
      affected memory related fields will return N/A and a suitable entry to this effect will be logged in opReports.log.

opReports 3.5.1 and newer include an improved percentile function

opReports 3.5.1, and newer ,now use the nearest rank inclusive percentile formula as provided in wikipedia:
https://en.wikipedia.org/wiki/Percentile

The differences between the new and old percentile formula cause very few differences in the 95th percentile outputs,
and when there are differences, they are generally very small fractional changes.