Skip to main content

NetApp_Insight_2020.png 

NetApp Knowledgebase

How are the blockSpace thresholds calculated for Element

Views:
193
Visibility:
Public
Votes:
0
Category:
element-software
Specialty:
solidfire
Last Updated:

 

Applies to

  • NetApp SolidFire SF-Series
  • NetApp H-Series

Answer

They are calculated based on the following equations:

1. Total block capacity - (largest node capacity + 3 % of total block capacity) = Cluster error threshold
2. Cluster error threshold / Total block capacity = % at which the system generates the error message

Additional Information

Block capacity is obtained by API Call GetClusterCapacity  under maxUsedSpace

You can consult raw capacity per node type on following KB: What is the raw block and metadata capacity for SolidFire and H-Series nodes?

Exmaple on how it is calculated:

For five SF9605 nodes:

  • Total block capacity of five SF9605 nodes (5 x 8,640GB) = 43,200 GB
  • Largest node capacity = 8,640GB

Three percent of total block capacity (43,200GB x 0.03) = 1,296 GB
Applying these values to the formulae, the following results are obtained:

1. Total block capacity - (largest node capacity + 3 % of total block capacity) = Cluster error threshold
2. Cluster error threshold / Total block capacity = % at which the system generates the error message 


43,200 - (8,640 + 1,296) = 33,264
33,264 / 43,200 = .77 (or 77 %)

Interpret the results of the above example as follows:

  • A BlockClusterFull error code will appear when the cluster is 77% (33,264 GB) full.
  • The system default for generating a warning severity message is 3% below the calculated error severity. In this example, the system would generate a warning severity when the cluster is 74% full.
  • If you want to be notified with a warning severity message before the 74% threshold is reached, you can specify the threshold percentage at which you want to see a warning on Cluster -> Settings -> Cluster Full Settings field.
  • For example, if you want a warning message at 10% below the error severity (77%), then enter 10 in the 'Raise a warning alert when _ % capacity remains before Helix could not recover' field. A warning severity will then be generated when 67% of the cluster is full.