koordlet: add libpfm4&perf group

[bowen-intel]

Ⅰ. Describe what this PR does

This pull request enhances the precision of CPI calculation through the utilization of the perf group. Additionally, it introduces the integration of libpfm4 as an event parser, thereby expanding the potential to incorporate additional TMA analyses in subsequent developments.

Ⅱ. Does this pull request fix one issue?

fixes #1476 #1470

Ⅲ. Describe how to verify it

Use Prometheus.

Ⅳ. Special notes for reviews

This pr uses unsafe.Pointer to mock syscall.Syscall6 in unit test, so I have to add unsafeptr tag for go vet tool and ignore github.com/koordinator-sh/koordinator/pkg/koordlet/util/perf package in test since it use -race tag.
Any comments is welcome.

V. Checklist

• I have written necessary docs and comments
• I have added necessary unit tests and integration tests
• All checks passed in make test

[codecov]

Codecov Report

Patch coverage: 55.83% and project coverage change: +0.53% 🎉

Comparison is base (3fcac00) 65.19% compared to head (b24b776) 65.72%.
Report is 53 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1554      +/-   ##
==========================================
+ Coverage   65.19%   65.72%   +0.53%     
==========================================
  Files         352      376      +24     
  Lines       36326    39842    +3516     
==========================================
+ Hits        23681    26186    +2505     
- Misses      10901    11745     +844     
- Partials     1744     1911     +167     

Flag	Coverage Δ
unittests	65.72% <55.83%> (+0.53%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
pkg/features/koordlet_features.go	100.00% <ø> (ø)
pkg/koordlet/util/perf/perf_linux.go	46.73% <0.00%> (ø)
...lectors/performance/performance_collector_linux.go	52.83% <21.42%> (-3.17%)	⬇️
pkg/koordlet/util/stat.go	52.85% <45.00%> (-22.15%)	⬇️
pkg/koordlet/util/perf_group/perf_group_linux.go	62.84% <62.84%> (ø)

... and 99 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[bowen-intel]

Make a A/B test for two method:

workload

100 nginx

apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-nginx
  namespace: nginx
  labels:
    run: my-nginx
spec:
  replicas: 100
  selector:
    matchLabels:
      run: my-nginx
  template:
    metadata:
      labels:
        run: my-nginx
    spec:
      nodeName: inspur-icx-1
      containers:
        - name: my-nginx
          image: nginx:latest
          ports:
            - containerPort: 80
          resources:
            requests:
              cpu: "1"
              memory: 1Gi
            limits:
              cpu: "1"
              memory: 1Gi
---
apiVersion: v1
kind: Service
metadata:
  name: my-nginx
  namespace: nginx
  labels:
    run: my-nginx
spec:
  ports:
    - port: 80
      protocol: TCP
      targetPort: 80
  selector:
    run: my-nginx

environment

CPU Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz
Memory 186Gi
Ubuntu 22.04
containerd github.com/containerd/containerd v1.6.8 9cd3357b7fd7218e4aec3eae239db1f68a5a6ec6
runc version 1.1.4
koordinator(based on v1.3.0)
kubernetes v1.27.2

libpfm4

perf-util

[bowen-intel]

Any comments and questions is welcome! 😀 @zwzhang0107 @songtao98

[songtao98]

/lgtm

[zwzhang0107]

Make a A/B test for two method:

workload

100 nginx

apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-nginx
  namespace: nginx
  labels:
    run: my-nginx
spec:
  replicas: 100
  selector:
    matchLabels:
      run: my-nginx
  template:
    metadata:
      labels:
        run: my-nginx
    spec:
      nodeName: inspur-icx-1
      containers:
        - name: my-nginx
          image: nginx:latest
          ports:
            - containerPort: 80
          resources:
            requests:
              cpu: "1"
              memory: 1Gi
            limits:
              cpu: "1"
              memory: 1Gi
---
apiVersion: v1
kind: Service
metadata:
  name: my-nginx
  namespace: nginx
  labels:
    run: my-nginx
spec:
  ports:
    - port: 80
      protocol: TCP
      targetPort: 80
  selector:
    run: my-nginx

environment

CPU Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz Memory 186Gi Ubuntu 22.04 containerd github.com/containerd/containerd v1.6.8 9cd3357b7fd7218e4aec3eae239db1f68a5a6ec6 runc version 1.1.4 koordinator(based on v1.3.0) kubernetes v1.27.2

libpfm4

perf-util

this looks great!

it will be better if we have the comparison on accuracy between perf and libpfm4

[bowen-intel]

Make a A/B test for two method:

workload

100 nginx

apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-nginx
  namespace: nginx
  labels:
    run: my-nginx
spec:
  replicas: 100
  selector:
    matchLabels:
      run: my-nginx
  template:
    metadata:
      labels:
        run: my-nginx
    spec:
      nodeName: inspur-icx-1
      containers:
        - name: my-nginx
          image: nginx:latest
          ports:
            - containerPort: 80
          resources:
            requests:
              cpu: "1"
              memory: 1Gi
            limits:
              cpu: "1"
              memory: 1Gi
---
apiVersion: v1
kind: Service
metadata:
  name: my-nginx
  namespace: nginx
  labels:
    run: my-nginx
spec:
  ports:
    - port: 80
      protocol: TCP
      targetPort: 80
  selector:
    run: my-nginx

environment

CPU Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz Memory 186Gi Ubuntu 22.04 containerd github.com/containerd/containerd v1.6.8 9cd3357b7fd7218e4aec3eae239db1f68a5a6ec6 runc version 1.1.4 koordinator(based on v1.3.0) kubernetes v1.27.2

libpfm4

perf-util

this looks great!

it will be better if we have the comparison on accuracy between perf and libpfm4

libpfm4

cycles: 9694773088 instructions: 2817277063

perf

perf stat -e '{cycles,instructions}' -G xxx.scope -a sleep 10

[zwzhang0107]

/lgtm

[koordinator-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: hormes

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [hormes]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment