[vc subcmd] init nodepool and workload subcommands

[fuweid]

The vc sub-command is used to setup simulated cluster over real cluster and simulate pod or node's lifecycle.
This sub-command uses kwok to setup virtual nodes.

Ideally, the user can use existing benchmark tools, like ArgoWorkflow, to setup large scale objects in the cluster.

1. nodepool

1.1 kperf vc nodepool add name [options]

This command uses statefulset to setup a set of virtual nodes.

Options:

• --nodes: optional, INT, default: 10, the number of virtual nodes
• --cpu: optional, INT, default: 8, the allocatable CPU resource per node
• --mem: optional, INT, default: 16, the allocatable Memory resource per node (GiB)
• --affinity: optional, STRING SLICE, default: empty slice, format: key=value1,value2,..., deploy nodepool controllers into nodes with specific labels.
• --max-pods: optional, INT, default: 110, the max running pods per virtual node.

TODO:

• [helmcli] support YAMLValuesApplier helper function #25
• [helmcli] support to show debug log #26
• [vc/nodepool subcmd] support --affinity for add operation #29
• [vc/nodepool subcmd] support --max-pods flag in add operation #30

1.2 kperf vc nodepool del name

This command is to delete virtual nodepool.

• [vc/nodepool subcmd] support delete operation #27

1.3 kperf vc nodepool list

This command is to show existing nodepool.

• [vc/nodepool subcmd] support list operation #28

2. workload

With running virtual-nodes, users can deploy workloads with that following nodeAffinity.

affinity:
   nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
         nodeSelectorTerms:
         - matchExpressions:
            - key: type
              operator: In
              values:
                - kperf-virtualnodes
            - key: alpha.kperf.io/nodepool
              operator: In
              values:
                 - {{ nodepool name }}

The workload command is a collection of common workloads to mutate kubernetes cluster.

TODO:

• [vc/workload subcmd] init job subcommand #31

NOTE: We can introduce Argo-Workflow/Kyverno/... workloads into vc/workload. But for this issue, we just need one subcommand.

Tasks

Preview Give feedback

1. cmd: init virtualcluster subcommand #7
   +0
2. *: init kperf-vc-nodepool add subcommand #17
   +0
3. [helmcli] support YAMLValuesApplier helper function #25
   kind/feature priority/P2 +2
   
4. [helmcli] support to show debug log #26
   kind/feature priority/P3 +2
   
5. [vc/nodepool subcmd] support delete operation #27
   kind/feature priority/P2 +2
   
6. [vc/nodepool subcmd] support list operation #28
   kind/feature priority/P2 +2
   
7. [vc/nodepool subcmd] support --affinity for add operation #29
   kind/feature priority/P3 +2
   
8. [vc/nodepool subcmd] support --max-pods flag in add operation #30
   kind/feature priority/P3 +2
   
9. [vc/workload subcmd] init job subcommand #31
   kind/feature priority/P2 +2

[fuweid]

Handled in contrib package like

• *: init node100_dp5_pod10k for runkperf bench  #117
• contrib: init bench subcommand to run test case #108

Closed it