HPA Beijing Testing

TESTING PROGRESS: 95%

Pairwise:

Dependencies: OOF, Multicloud, SO, Policy, AAI

 HPA is unique in that it relies on a multitude of projects and OOF is new and untested. Most of HPA work is gated by OOF, but we are doing our best to help facilitate OOF testing.

Updates: 

Thursday     5/31/18

Tuesday     5/29/18

Wednesday     5/23/18

Tuesday     5/22/18 

Thursday     5/17/18 

Wednesday      5/16/18 

Tuesday      5/15/18

Monday      5/14/18

Friday      5/11/18

Integration:

vCPE Regression (ensure HPA changes do not break vCPE)

• Kang is working through this

vCPE + HPA

• Gated by vCPE Regression, If vCPE doesn’t work in regression it won’t work with HPA

• The work to test HPA over vCPE regression is only + 10%

vCPE + HPA Integration Test Plan

• Setup vCPE the same as regression; up to Service Instantiation

• Before vCPE Service Instantiation

  • Create HPA policies – See wiki page Policy OOF HPA 
    Example policies: 

    HPA Policy Example 1

    { "riskLevel":"1", "riskType":"SampleRiskType", "policyName":"PolicyHPA3", "service":"hpaPolicy", "guard":"False", "description":"Must have Huge pages support", "templateVersion":"OpenSource.version.1", "priority":"1", "version":"CSIT", "content":{ "identity":"hpaPolicy_set3", "policyScope":[ "INTERNATIONAL" ], "policyType":"hpaPolicy", "resources":"vGMuxInfra", "flavorFeatures":[ { "flavorLabel":"vcpe.vgmux", "flavorProperties":[ { "hpa-feature-attributes":[ { "hpa-attribute-key":"numVirtualCpu", "hpa-attribute-value":"4", "operator":"=" }, { "hpa-attribute-key":"virtualMemSize", "hpa-attribute-value":"8192", "operator":"=", "unit":"MB" } ], "mandatory":"True", "hpa-feature":"basicCapabilities", "architecture":"generic", "hpa-version":"v1" }, { "hpa-feature-attributes":[ { "hpa-attribute-key":"logicalCpuPinningPolicy", "hpa-attribute-value":"dedicated", "operator":"=" } ], "mandatory":"True", "hpa-feature":"cpuPinning", "architecture":"generic", "hpa-version":"v1", } ] } ] } }

    HPA Policy Example 2

    { "riskLevel":"1", "riskType":"SampleRiskType", "policyName":"PolicyHPA3", "service":"hpaPolicy", "guard":"False", "description":"Must have Huge pages support", "templateVersion":"OpenSource.version.1", "priority":"1", "version":"CSIT", "content":{ "identity":"hpaPolicy_set3", "policyScope":[ "INTERNATIONAL" ], "policyType":"hpaPolicy", "resources":"vGMuxInfra", "flavorFeatures":[ { "flavorLabel":"vcpe.vgmux", "flavorProperties":[ { "hpa-feature-attributes":[ { "hpa-attribute-key":"numVirtualCpu", "hpa-attribute-value":"4", "operator":">=" }, { "hpa-attribute-key":"virtualMemSize", "hpa-attribute-value":"8", "operator":"=", "unit":"MB" } ], "mandatory":"True", "hpa-feature":"basicCapabilities", "architecture":"generic", "hpa-version":"v1" } ] } ] } }

    
     This example will change based on flavors in OpenStack/Multicloud in the lab.

  • Create flavors in OpenStack

    • flavors which is used to discover HPA information should named with prefix of "onap." , otherwise there will no HPA information can be extracted by multicloud plugins for OpenStack

    • the number of flavors to be created for ONAP is determined by number of the generic flavors multiplied by the combination of HPA specification.

      • In this example, there are 2 generic flavors:

        • medium: 4 vcpu, 4GB memory, 40GB storage

        • large: 6 vcpu, 8GB memory, 80GB storage

      • In this example, there are 2 set of HPA specification:
        
        

        • set 1: hw:numa_nodes=2, hw:cpu_policy=dedicated, hw:mem_page_size=2M;

        • set 2: hw:cpu_policy=dedicated, hw:mem_page_size=2M

      • Hence the flavors named with prefix of "onap." are:

        • onap.flavor2.medium

          • medium with HPA set 1

        • onap.flavor2.large

          • large  with HPA set 1

        • onap.flavor3.medium

          • medium with HPA set 2

        • onap.flavor3.large

          • large with HPA set 2
            
            
            

  • Use modified CSARs flavorLabel attributes to match the ones created in policy - See below 'HPA Updated CSAR's' section.

  • Distribute CSARs to SO

• At vCPE Service Instantiation

  • Send updated REST service instantiation request to SO that includes user param: Customer_Location

• Continue the same as vCPE Regression and confirm all is well

Updates:

Thursday     5/31/18

Tuesday     5/29/18

Wednesday      5/16/18 

HPA Updated CSAR's

csars_20180516_HPA.zip

The above CSAR's are copies of the ones Kang is using for vCPE Regression Integration with one minor change.

Each CSAR has heat Environment File update to vcpe_flavor_name as follows:

HPA Code uses the vcpe_flavor_name to match HPA flavors to the correct policy and service. 

Postman Collection

Associate a Complex Object with a cloud region

Associate Complex with Cloud Region

Another way to populate the default cloud region with complex is the robot script:

    Attach to robot VM, execute following script:

       $ /opt/demo.sh init_customer

    The default cloud-region (CloudOwner,RegionOne) and the default complex (clli1) and the association between them will be populated .

    Caveat:  The cloud-region populated by this script is not conform to the cloud region expected by multicloud. So it is not possible to discover/populate HPA information under this default cloud region.