update deps

Signed-off-by: Vasiliy Tolstov <v.tolstov@unistack.org>

.github/dependabot.yml

@@ -1,19 +0,0 @@
-# To get started with Dependabot version updates, you'll need to specify which
-# package ecosystems to update and where the package manifests are located.
-# Please see the documentation for all configuration options:
-# https://help.github.com/github/administering-a-repository/configuration-options-for-dependency-updates
-
-version: 2
-updates:
-
-  # Maintain dependencies for GitHub Actions
-  - package-ecosystem: "github-actions"
-    directory: "/"
-    schedule:
-      interval: "daily"
-
-  # Maintain dependencies for Golang
-  - package-ecosystem: "gomod"
-    directory: "/"
-    schedule:
-      interval: "daily"

.github/workflows/build.yml

@@ -1,46 +0,0 @@
-name: build
-on:
- push:
-    branches:
-    - master
-jobs:
-  test:
-    name: test
-    runs-on: ubuntu-latest
-    steps:
-    - name: setup
-      uses: actions/setup-go@v2
-      with:
-        go-version: 1.16
-    - name: checkout
-      uses: actions/checkout@v2
-    - name: cache
-      uses: actions/cache@v2
-      with:
-        path: ~/go/pkg/mod
-        key: ${{ runner.os }}-go-${{ hashFiles('**/go.sum') }}
-        restore-keys: ${{ runner.os }}-go-
-    - name: deps
-      run: go get -v -t -d ./...
-    - name: test
-      env:
-        INTEGRATION_TESTS: yes
-      run: go test -mod readonly -v ./...
-  lint:
-    name: lint
-    runs-on: ubuntu-latest
-    steps:
-      - name: checkout
-        uses: actions/checkout@v2
-      - name: lint
-        uses: golangci/golangci-lint-action@v2
-        continue-on-error: true
-        with:
-          # Required: the version of golangci-lint is required and must be specified without patch version: we always use the latest patch version.
-          version: v1.30
-          # Optional: working directory, useful for monorepos
-          # working-directory: somedir
-          # Optional: golangci-lint command line arguments.
-          # args: --issues-exit-code=0
-          # Optional: show only new issues if it's a pull request. The default value is `false`.
-          # only-new-issues: true

.github/workflows/codeql-analysis.yml

@@ -1,75 +0,0 @@
-# For most projects, this workflow file will not need changing; you simply need
-# to commit it to your repository.
-#
-# You may wish to alter this file to override the set of languages analyzed,
-# or to provide custom queries or build logic.
-#
-# ******** NOTE ********
-# We have attempted to detect the languages in your repository. Please check
-# the `language` matrix defined below to confirm you have the correct set of
-# supported CodeQL languages.
-#
-name: "CodeQL"
-
-on:
-  workflow_run:
-    workflows: ["prbuild"]
-    types:
-      - completed
-  push:
-    branches: [ master ]
-  pull_request:
-    # The branches below must be a subset of the branches above
-    branches: [ master ]
-  schedule:
-    - cron: '34 1 * * 0'
-
-jobs:
-  analyze:
-    name: Analyze
-    runs-on: ubuntu-latest
-    permissions:
-      actions: read
-      contents: read
-      security-events: write
-
-    strategy:
-      fail-fast: false
-      matrix:
-        language: [ 'go' ]
-        # CodeQL supports [ 'cpp', 'csharp', 'go', 'java', 'javascript', 'python' ]
-        # Learn more:
-        # https://docs.github.com/en/free-pro-team@latest/github/finding-security-vulnerabilities-and-errors-in-your-code/configuring-code-scanning#changing-the-languages-that-are-analyzed
-
-    steps:
-    - name: Checkout repository
-      uses: actions/checkout@v2
-
-    # Initializes the CodeQL tools for scanning.
-    - name: Initialize CodeQL
-      uses: github/codeql-action/init@v1
-      with:
-        languages: ${{ matrix.language }}
-        # If you wish to specify custom queries, you can do so here or in a config file.
-        # By default, queries listed here will override any specified in a config file.
-        # Prefix the list here with "+" to use these queries and those in the config file.
-        # queries: ./path/to/local/query, your-org/your-repo/queries@main
-
-    # Autobuild attempts to build any compiled languages  (C/C++, C#, or Java).
-    # If this step fails, then you should remove it and run the build manually (see below)
-    - name: Autobuild
-      uses: github/codeql-action/autobuild@v1
-
-    # ℹ️ Command-line programs to run using the OS shell.
-    # 📚 https://git.io/JvXDl
-
-    # ✏️ If the Autobuild fails above, remove it and uncomment the following three lines
-    #    and modify them (or add more) to build your code if your project
-    #    uses a compiled language
-
-    #- run: |
-    #   make bootstrap
-    #   make release
-
-    - name: Perform CodeQL Analysis
-      uses: github/codeql-action/analyze@v1

.github/workflows/dependabot-automerge.yml

@@ -1,66 +0,0 @@
-name: "prautomerge"
-
-on:
-  workflow_run:
-    workflows: ["prbuild"]
-    types:
-      - completed
-
-permissions:
-  contents: write
-  pull-requests: write
-
-jobs:
-  Dependabot-Automerge:
-    runs-on: ubuntu-latest
-    # Contains workaround to execute if dependabot updates the PR by checking for the base branch in the linked PR
-    # The the github.event.workflow_run.event value is 'push' and not 'pull_request'
-    # dont work with multiple workflows when last returns success
-    if: >-
-      github.event.workflow_run.conclusion == 'success'
-      && github.actor == 'dependabot[bot]'
-      && github.event.sender.login == 'dependabot[bot]'
-      && github.event.sender.type == 'Bot'
-      && (github.event.workflow_run.event == 'pull_request'
-          || (github.event.workflow_run.event == 'push' && github.event.workflow_run.pull_requests[0].base.ref == github.event.repository.default_branch ))
-    steps:
-      - name: Approve Changes and Merge changes if label 'dependencies' is set
-        uses: actions/github-script@v5
-        with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
-          script: |
-            console.log(context.payload.workflow_run);
-            
-            var labelNames = await github.paginate(
-              github.issues.listLabelsOnIssue,
-              {
-                repo: context.repo.repo,
-                owner: context.repo.owner,
-                issue_number: context.payload.workflow_run.pull_requests[0].number,
-              },
-              (response) => response.data.map(
-                (label) => label.name
-              )
-            );
-
-            console.log(labelNames);
-
-            if (labelNames.includes('dependencies')) {
-              console.log('Found label');
-
-              await github.pulls.createReview({
-                repo: context.repo.repo,
-                owner: context.repo.owner,
-                pull_number: context.payload.workflow_run.pull_requests[0].number,
-                event: 'APPROVE'
-              });
-              console.log('Approved PR');
-
-              await github.pulls.merge({
-                repo: context.repo.repo,
-                owner: context.repo.owner,
-                pull_number: context.payload.workflow_run.pull_requests[0].number,
-              });
-
-              console.log('Merged PR');
-            }

.github/workflows/job_coverage.yml

@@ -0,0 +1,53 @@
+name: coverage
+
+on:
+  push:
+    branches: [ main, v3, v4 ]
+    paths-ignore:
+      - '.github/**'
+      - '.gitea/**'
+  pull_request:
+    branches: [ main, v3, v4 ]
+
+jobs:
+
+  build:
+    if: github.server_url != 'https://github.com'
+    runs-on: ubuntu-latest
+    steps:
+    - name: checkout code
+      uses: actions/checkout@v4
+      with:
+        filter: 'blob:none'
+
+    - name: setup go
+      uses: actions/setup-go@v5
+      with:
+        cache-dependency-path: "**/*.sum"
+        go-version: 'stable'
+
+    - name: test coverage
+      run: |
+        go test -v -cover ./... -covermode=count -coverprofile coverage.out -coverpkg ./...
+        go tool cover -func coverage.out -o coverage.out
+
+    - name: coverage badge
+      uses: tj-actions/coverage-badge-go@v2
+      with:
+        green: 80
+        filename: coverage.out
+
+    - uses: stefanzweifel/git-auto-commit-action@v4
+      name: autocommit
+      with:
+        commit_message: Apply Code Coverage Badge
+        skip_fetch: false
+        skip_checkout: false
+        file_pattern: ./README.md
+
+    - name: push
+      if: steps.auto-commit-action.outputs.changes_detected == 'true'
+      uses: ad-m/github-push-action@master
+      with:
+        github_token: ${{ github.token }}
+        branch: ${{ github.ref }}

.github/workflows/job_lint.yml

@@ -0,0 +1,29 @@
+name: lint
+
+on:
+  pull_request:
+    types: [opened, reopened, synchronize]
+    branches: [ master, v3, v4 ]
+    paths-ignore:
+      - '.github/**'
+      - '.gitea/**'
+
+jobs:
+  lint:
+    runs-on: ubuntu-latest
+    steps:
+    - name: checkout code
+      uses: actions/checkout@v4
+      with:
+        filter: 'blob:none'
+    - name: setup go
+      uses: actions/setup-go@v5
+      with:
+        cache-dependency-path: "**/*.sum"
+        go-version: 'stable'
+    - name: setup deps
+      run: go get -v ./...
+    - name: run lint
+      uses: golangci/golangci-lint-action@v6
+      with:
+        version: 'latest'

.github/workflows/job_test.yml

@@ -0,0 +1,31 @@
+name: test
+
+on:
+  pull_request:
+    types: [opened, reopened, synchronize]
+    branches: [ master, v3, v4 ]
+  push:
+    branches: [ master, v3, v4 ]
+    paths-ignore:
+      - '.github/**'
+      - '.gitea/**'
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+    - name: checkout code
+      uses: actions/checkout@v4
+      with:
+        filter: 'blob:none'
+    - name: setup go
+      uses: actions/setup-go@v5
+      with:
+        cache-dependency-path: "**/*.sum"
+        go-version: 'stable'
+    - name: setup deps
+      run: go get -v ./...
+    - name: run test
+      env:
+        INTEGRATION_TESTS: yes
+      run: go test -mod readonly -v ./...

.github/workflows/job_tests.yml

@@ -0,0 +1,50 @@
+name: test
+
+on:
+  pull_request:
+    types: [opened, reopened, synchronize]
+    branches: [ master, v3, v4 ]
+  push:
+    branches: [ master, v3, v4 ]
+    paths-ignore:
+      - '.github/**'
+      - '.gitea/**'
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+    - name: checkout code
+      uses: actions/checkout@v4
+      with:
+        filter: 'blob:none'
+    - name: checkout tests
+      uses: actions/checkout@v4
+      with:
+        ref: master
+        filter: 'blob:none'
+        repository: unistack-org/micro-tests
+        path: micro-tests
+    - name: setup go
+      uses: actions/setup-go@v5
+      with:
+        cache-dependency-path: "**/*.sum"
+        go-version: 'stable'
+    - name: setup go work
+      env:
+        GOWORK: ${{ github.workspace }}/go.work
+      run: |
+        go work init
+        go work use .
+        go work use micro-tests
+    - name: setup deps
+      env:
+        GOWORK: ${{ github.workspace }}/go.work
+      run: go get -v ./...
+    - name: run tests
+      env:
+        INTEGRATION_TESTS: yes
+        GOWORK: ${{ github.workspace }}/go.work
+      run: |
+        cd micro-tests
+        go test -mod readonly -v ./... || true

.github/workflows/pr.yml

@@ -1,46 +0,0 @@
-name: prbuild
-on:
-  pull_request:
-    branches:
-    - master
-jobs:
-  test:
-    name: test
-    runs-on: ubuntu-latest
-    steps:
-    - name: setup
-      uses: actions/setup-go@v2
-      with:
-        go-version: 1.16
-    - name: checkout
-      uses: actions/checkout@v2
-    - name: cache
-      uses: actions/cache@v2
-      with:
-        path: ~/go/pkg/mod
-        key: ${{ runner.os }}-go-${{ hashFiles('**/go.sum') }}
-        restore-keys: ${{ runner.os }}-go-
-    - name: deps
-      run: go get -v -t -d ./...
-    - name: test
-      env:
-        INTEGRATION_TESTS: yes
-      run: go test -mod readonly -v ./...
-  lint:
-    name: lint
-    runs-on: ubuntu-latest
-    steps:
-      - name: checkout
-        uses: actions/checkout@v2
-      - name: lint
-        uses: golangci/golangci-lint-action@v2
-        continue-on-error: true
-        with:
-          # Required: the version of golangci-lint is required and must be specified without patch version: we always use the latest patch version.
-          version: v1.30
-          # Optional: working directory, useful for monorepos
-          # working-directory: somedir
-          # Optional: golangci-lint command line arguments.
-          # args: --issues-exit-code=0
-          # Optional: show only new issues if it's a pull request. The default value is `false`.
-          # only-new-issues: true

.gitignore

@@ -13,3 +13,9 @@
 
 # Dependency directories (remove the comment below to include it)
 # vendor/
+
+# General
+.DS_Store
+.idea
+.vscode
+bin/

.golangci.yml

@@ -0,0 +1,5 @@
+run:
+  concurrency: 8
+  timeout: 5m
+  issues-exit-code: 1
+  tests: true

README.md

@@ -1,2 +1,9 @@
-# micro-broker-kgo
-yet another micro kafka broker alternative
+# broker-kgo
+
+TODO:
+
+1) экспортируем текущий оффсет для каждой партиции в группе
+2) экспортируем лаг для группы
+3) мониторим
+   1) если есть лаг больше нуля
+   2) если дельта оффсета за нужное нам время не 

carrier.go

@@ -0,0 +1,93 @@
+package kgo
+
+import (
+	"net/http"
+	"slices"
+	"strings"
+
+	"github.com/twmb/franz-go/pkg/kgo"
+	"go.unistack.org/micro/v3/metadata"
+)
+
+// RecordCarrier injects and extracts traces from a kgo.Record.
+//
+// This type exists to satisfy the otel/propagation.TextMapCarrier interface.
+type RecordCarrier struct {
+	record *kgo.Record
+}
+
+// NewRecordCarrier creates a new RecordCarrier.
+func NewRecordCarrier(record *kgo.Record) RecordCarrier {
+	return RecordCarrier{record: record}
+}
+
+// Get retrieves a single value for a given key if it exists.
+func (c RecordCarrier) Get(key string) string {
+	for _, h := range c.record.Headers {
+		if h.Key == key {
+			return string(h.Value)
+		}
+	}
+	return ""
+}
+
+// Set sets a header.
+func (c RecordCarrier) Set(key, val string) {
+	// Check if key already exists.
+	for i, h := range c.record.Headers {
+		if h.Key == key {
+			// Key exist, update the value.
+			c.record.Headers[i].Value = []byte(val)
+			return
+		}
+	}
+	// Key does not exist, append new header.
+	c.record.Headers = append(c.record.Headers, kgo.RecordHeader{
+		Key:   key,
+		Value: []byte(val),
+	})
+}
+
+// Keys returns a slice of all key identifiers in the carrier.
+func (c RecordCarrier) Keys() []string {
+	out := make([]string, len(c.record.Headers))
+	for i, h := range c.record.Headers {
+		out[i] = h.Key
+	}
+	return out
+}
+
+func setHeaders(r *kgo.Record, md metadata.Metadata, exclude ...string) {
+	seen := make(map[string]struct{})
+
+loop:
+	for k, v := range md {
+		k = http.CanonicalHeaderKey(k)
+
+		if _, ok := seen[k]; ok {
+			continue loop
+		}
+
+		if slices.ContainsFunc(exclude, func(s string) bool {
+			return strings.EqualFold(s, k)
+		}) {
+			continue loop
+		}
+
+		for i := 0; i < len(r.Headers); i++ {
+			if strings.EqualFold(r.Headers[i].Key, k) {
+				// Key exist, update the value.
+				r.Headers[i].Value = []byte(v)
+				continue loop
+			}
+		}
+
+		// Key does not exist, append new header.
+		r.Headers = append(r.Headers, kgo.RecordHeader{
+			Key:   k,
+			Value: []byte(v),
+		})
+
+		seen[k] = struct{}{}
+	}
+}

errors.go

@@ -0,0 +1,10 @@
+package kgo
+
+import (
+	"context"
+	"errors"
+)
+
+func isContextError(err error) bool {
+	return errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded)
+}

event.go

@@ -0,0 +1,50 @@
+package kgo
+
+import (
+	"context"
+	"sync"
+
+	"go.unistack.org/micro/v3/broker"
+)
+
+type event struct {
+	msg *broker.Message
+	err error
+	ctx context.Context
+
+	topic string
+
+	sync.RWMutex
+	ack bool
+}
+
+func (p *event) Context() context.Context {
+	return p.ctx
+}
+
+func (p *event) Topic() string {
+	return p.topic
+}
+
+func (p *event) Message() *broker.Message {
+	return p.msg
+}
+
+func (p *event) Ack() error {
+	p.ack = true
+	return nil
+}
+
+func (p *event) Error() error {
+	return p.err
+}
+
+func (p *event) SetError(err error) {
+	p.err = err
+}
+
+var eventPool = sync.Pool{
+	New: func() interface{} {
+		return &event{msg: &broker.Message{}}
+	},
+}

go.mod

@@ -1,10 +1,32 @@
 module go.unistack.org/micro-broker-kgo/v3
 
-go 1.16
+go 1.24.0
+
+toolchain go1.24.3
 
 require (
-	github.com/klauspost/compress v1.13.6 // indirect
-	github.com/twmb/franz-go v1.1.2
-	github.com/twmb/franz-go/pkg/kmsg v0.0.0-20210930203137-7fbfd17de279
-	go.unistack.org/micro/v3 v3.8.0
+	github.com/google/uuid v1.6.0
+	github.com/stretchr/testify v1.11.1
+	github.com/twmb/franz-go v1.20.2
+	github.com/twmb/franz-go/pkg/kadm v1.17.1
+	github.com/twmb/franz-go/pkg/kmsg v1.12.0
+	go.opentelemetry.io/otel v1.38.0
+	go.unistack.org/micro/v3 v3.11.48
+)
+
+require (
+	github.com/ash3in/uuidv8 v1.2.0 // indirect
+	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect
+	github.com/klauspost/compress v1.18.1 // indirect
+	github.com/kr/pretty v0.3.1 // indirect
+	github.com/matoous/go-nanoid v1.5.1 // indirect
+	github.com/pierrec/lz4/v4 v4.1.22 // indirect
+	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
+	go.unistack.org/micro-proto/v3 v3.4.1 // indirect
+	golang.org/x/crypto v0.43.0 // indirect
+	golang.org/x/sys v0.37.0 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20251022142026-3a174f9686a8 // indirect
+	google.golang.org/grpc v1.76.0 // indirect
+	google.golang.org/protobuf v1.36.10 // indirect
+	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -1,67 +1,74 @@
-github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/ef-ds/deque v1.0.4/go.mod h1:gXDnTC3yqvBcHbq2lcExjtAcVrOnJCbMcZXmuj8Z4tg=
-github.com/golang-jwt/jwt/v4 v4.1.0/go.mod h1:/xlHOz8bRuivTWchD4jCa+NbatV+wEUSzwAxVc6locg=
-github.com/golang/protobuf v1.5.0/go.mod h1:FsONVRAS9T7sI+LIUmWTfcYkHO4aIWwzhcaSAoJOfIk=
-github.com/google/go-cmp v0.3.1/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMywk6iLU=
-github.com/google/go-cmp v0.5.5 h1:Khx7svrCpmxxtHBq5j2mp/xVjsi8hQMfNLvJFAlrGgU=
-github.com/google/go-cmp v0.5.5/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/gorilla/securecookie v1.1.1/go.mod h1:ra0sb63/xPlUeL+yeDciTfxMRAA+MP+HVt/4epWDjd4=
-github.com/gorilla/sessions v1.2.1/go.mod h1:dk2InVEVJ0sfLlnXv9EAgkf6ecYs/i80K/zI+bUmuGM=
-github.com/hashicorp/go-uuid v1.0.2/go.mod h1:6SBZvOh/SIDV7/2o3Jml5SYk/TvGqwFJ/bN7x4byOro=
-github.com/imdario/mergo v0.3.12/go.mod h1:jmQim1M+e3UYxmgPu/WyfjB3N3VflVyUjjjwH0dnCYA=
-github.com/jcmturner/aescts/v2 v2.0.0/go.mod h1:AiaICIRyfYg35RUkr8yESTqvSy7csK90qZ5xfvvsoNs=
-github.com/jcmturner/dnsutils/v2 v2.0.0/go.mod h1:b0TnjGOvI/n42bZa+hmXL+kFJZsFT7G4t3HTlQ184QM=
-github.com/jcmturner/gofork v1.0.0/go.mod h1:MK8+TM0La+2rjBD4jE12Kj1pCCxK7d2LK/UM3ncEo0o=
-github.com/jcmturner/goidentity/v6 v6.0.1/go.mod h1:X1YW3bgtvwAXju7V3LCIMpY0Gbxyjn/mY9zx4tFonSg=
-github.com/jcmturner/gokrb5/v8 v8.4.2/go.mod h1:sb+Xq/fTY5yktf/VxLsE3wlfPqQjp0aWNYyvBVK62bc=
-github.com/jcmturner/rpc/v2 v2.0.3/go.mod h1:VUJYCIDm3PVOEHw8sgt091/20OJjskO/YJki3ELg/Hc=
-github.com/klauspost/compress v1.13.5/go.mod h1:/3/Vjq9QcHkK5uEr5lBEmyoZ1iFhe47etQ6QUkpK6sk=
-github.com/klauspost/compress v1.13.6 h1:P76CopJELS0TiO2mebmnzgWaajssP/EszplttgQxcgc=
-github.com/klauspost/compress v1.13.6/go.mod h1:/3/Vjq9QcHkK5uEr5lBEmyoZ1iFhe47etQ6QUkpK6sk=
-github.com/patrickmn/go-cache v2.1.0+incompatible/go.mod h1:3Qf8kWWT7OJRJbdiICTKqZju1ZixQ/KpMGzzAfe6+WQ=
-github.com/pierrec/lz4/v4 v4.1.8 h1:ieHkV+i2BRzngO4Wd/3HGowuZStgq6QkPsD1eolNAO4=
-github.com/pierrec/lz4/v4 v4.1.8/go.mod h1:gZWDp/Ze/IJXGXf23ltt2EXimqmTUXEy0GFuRQyBid4=
-github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
-github.com/silas/dag v0.0.0-20210626123444-3804bac2d6d4/go.mod h1:7RTUFBdIRC9nZ7/3RyRNH1bdqIShrDejd1YbLwgPS+I=
-github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
-github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81PSLYec5m4=
-github.com/stretchr/testify v1.6.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
-github.com/twmb/franz-go v1.1.2 h1:i9lCDEcXWpRd3YXFlffIc/koBl+JijI13F5sg69NPRk=
-github.com/twmb/franz-go v1.1.2/go.mod h1:KerrVhzNpasYrWJLr2Yj6Cui43f1BxH4U9SJEDVOjqQ=
-github.com/twmb/franz-go/pkg/kmsg v0.0.0-20210914042331-106aef61b693/go.mod h1:SxG/xJKhgPu25SamAq0rrucfp7lbzCpEXOC+vH/ELrY=
-github.com/twmb/franz-go/pkg/kmsg v0.0.0-20210930203137-7fbfd17de279 h1:58GX0Ju/PCG1zWVdsLEE5OUu70KlfOrxY7djPszuqiw=
-github.com/twmb/franz-go/pkg/kmsg v0.0.0-20210930203137-7fbfd17de279/go.mod h1:SxG/xJKhgPu25SamAq0rrucfp7lbzCpEXOC+vH/ELrY=
-github.com/twmb/go-rbtree v1.0.0 h1:KxN7dXJ8XaZ4cvmHV1qqXTshxX3EBvX/toG5+UR49Mg=
-github.com/twmb/go-rbtree v1.0.0/go.mod h1:UlIAI8gu3KRPkXSobZnmJfVwCJgEhD/liWzT5ppzIyc=
-github.com/unistack-org/micro-proto v0.0.9 h1:KrWLS4FUX7UAWNAilQf70uad6ZPf/0EudeddCXllRVc=
-github.com/unistack-org/micro-proto v0.0.9/go.mod h1:Cckwmzd89gvS7ThxzZp9kQR/EOdksFQcsTAtDDyKwrg=
-go.unistack.org/micro/v3 v3.8.0 h1:9k6C40xdJf65VW6cVB8sOJVSJhyz1UfFOhX7/II2ivo=
-go.unistack.org/micro/v3 v3.8.0/go.mod h1:Tkteri0wiiybbH6aPqay26pZHFIAwL9LXJc2x1Jkakk=
-golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
-golang.org/x/crypto v0.0.0-20201112155050-0c6587e931a9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
-golang.org/x/crypto v0.0.0-20210817164053-32db794688a5/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc=
-golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
-golang.org/x/net v0.0.0-20200114155413-6afb5195e5aa/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
-golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.0.0-20210913180222-943fd674d43e/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.0.0-20210928044308-7d9f5e0b762b/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
-golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210423082822-04245dca01da/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
-golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
-golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
-golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1 h1:go1bK/D/BFZV2I8cIQd1NKEZ+0owSTG1fDTci4IqFcE=
-golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
-google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw=
-google.golang.org/protobuf v1.27.1 h1:SnqbnDw1V7RiZcXPx5MEeqPv2s79L9i7BJUlG/+RurQ=
-google.golang.org/protobuf v1.27.1/go.mod h1:9q0QmTI4eRPtz6boOQmLYwt+qCgq0jsYwAQnmE0givc=
+github.com/DATA-DOG/go-sqlmock v1.5.2 h1:OcvFkGmslmlZibjAjaHm3L//6LiuBgolP7OputlJIzU=
+github.com/DATA-DOG/go-sqlmock v1.5.2/go.mod h1:88MAG/4G7SMwSE3CeA0ZKzrT5CiOU3OJ+JlNzwDqpNU=
+github.com/ash3in/uuidv8 v1.2.0 h1:2oogGdtCPwaVtyvPPGin4TfZLtOGE5F+W++E880G6SI=
+github.com/ash3in/uuidv8 v1.2.0/go.mod h1:BnU0wJBxnzdEKmVg4xckBkD+VZuecTFTUP3M0dWgyY4=
+github.com/creack/pty v1.1.9/go.mod h1:oKZEueFk5CKHvIhNR5MUki03XCEU+Q6VDXinZuGJ33E=
+github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc h1:U9qPSI2PIWSS1VwoXQT9A3Wy9MM3WgvqSxFWenqJduM=
+github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek=
+github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
+github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
+github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
+github.com/klauspost/compress v1.18.0 h1:c/Cqfb0r+Yi+JtIEq73FWXVkRonBlf0CRNYc8Zttxdo=
+github.com/klauspost/compress v1.18.0/go.mod h1:2Pp+KzxcywXVXMr50+X0Q/Lsb43OQHYWRCY2AiWywWQ=
+github.com/klauspost/compress v1.18.1 h1:bcSGx7UbpBqMChDtsF28Lw6v/G94LPrrbMbdC3JH2co=
+github.com/klauspost/compress v1.18.1/go.mod h1:ZQFFVG+MdnR0P+l6wpXgIL4NTtwiKIdBnrBd8Nrxr+0=
+github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
+github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk=
+github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
+github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
+github.com/matoous/go-nanoid v1.5.1 h1:aCjdvTyO9LLnTIi0fgdXhOPPvOHjpXN6Ik9DaNjIct4=
+github.com/matoous/go-nanoid v1.5.1/go.mod h1:zyD2a71IubI24efhpvkJz+ZwfwagzgSO6UNiFsZKN7U=
+github.com/pierrec/lz4/v4 v4.1.22 h1:cKFw6uJDK+/gfw5BcDL0JL5aBsAFdsIT18eRtLj7VIU=
+github.com/pierrec/lz4/v4 v4.1.22/go.mod h1:gZWDp/Ze/IJXGXf23ltt2EXimqmTUXEy0GFuRQyBid4=
+github.com/pkg/diff v0.0.0-20210226163009-20ebb0f2a09e/go.mod h1:pJLUxLENpZxwdsKMEsNbx1VGcRFpLqf3715MtcvvzbA=
+github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRIccs7FGNTlIRMkT8wgtp5eCXdBlqhYGL6U=
+github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/rogpeppe/go-internal v1.9.0/go.mod h1:WtVeX8xhTBvf0smdhujwtBcq4Qrzq/fJaraNFVN+nFs=
+github.com/rogpeppe/go-internal v1.13.1 h1:KvO1DLK/DRN07sQ1LQKScxyZJuNnedQ5/wKSR38lUII=
+github.com/rogpeppe/go-internal v1.13.1/go.mod h1:uMEvuHeurkdAXX61udpOXGD/AzZDWNMNyH2VO9fmH0o=
+github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
+github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
+github.com/twmb/franz-go v1.19.5 h1:W7+o8D0RsQsedqib71OVlLeZ0zI6CbFra7yTYhZTs5Y=
+github.com/twmb/franz-go v1.19.5/go.mod h1:4kFJ5tmbbl7asgwAGVuyG1ZMx0NNpYk7EqflvWfPCpM=
+github.com/twmb/franz-go v1.20.2 h1:CiwhyKZHW6vqSHJkh+RTxFAJkio0jBjM/JQhx/HZ72A=
+github.com/twmb/franz-go v1.20.2/go.mod h1:YCnepDd4gl6vdzG03I5Wa57RnCTIC6DVEyMpDX/J8UA=
+github.com/twmb/franz-go/pkg/kadm v1.16.1 h1:IEkrhTljgLHJ0/hT/InhXGjPdmWfFvxp7o/MR7vJ8cw=
+github.com/twmb/franz-go/pkg/kadm v1.16.1/go.mod h1:Ue/ye1cc9ipsQFg7udFbbGiFNzQMqiH73fGC2y0rwyc=
+github.com/twmb/franz-go/pkg/kadm v1.17.1 h1:Bt02Y/RLgnFO2NP2HVP1kd2TFtGRiJZx+fSArjZDtpw=
+github.com/twmb/franz-go/pkg/kadm v1.17.1/go.mod h1:s4duQmrDbloVW9QTMXhs6mViTepze7JLG43xwPcAeTg=
+github.com/twmb/franz-go/pkg/kmsg v1.11.2 h1:hIw75FpwcAjgeyfIGFqivAvwC5uNIOWRGvQgZhH4mhg=
+github.com/twmb/franz-go/pkg/kmsg v1.11.2/go.mod h1:CFfkkLysDNmukPYhGzuUcDtf46gQSqCZHMW1T4Z+wDE=
+github.com/twmb/franz-go/pkg/kmsg v1.12.0 h1:CbatD7ers1KzDNgJqPbKOq0Bz/WLBdsTH75wgzeVaPc=
+github.com/twmb/franz-go/pkg/kmsg v1.12.0/go.mod h1:+DPt4NC8RmI6hqb8G09+3giKObE6uD2Eya6CfqBpeJY=
+go.opentelemetry.io/otel v1.38.0 h1:RkfdswUDRimDg0m2Az18RKOsnI8UDzppJAtj01/Ymk8=
+go.opentelemetry.io/otel v1.38.0/go.mod h1:zcmtmQ1+YmQM9wrNsTGV/q/uyusom3P8RxwExxkZhjM=
+go.unistack.org/micro-proto/v3 v3.4.1 h1:UTjLSRz2YZuaHk9iSlVqqsA50JQNAEK2ZFboGqtEa9Q=
+go.unistack.org/micro-proto/v3 v3.4.1/go.mod h1:okx/cnOhzuCX0ggl/vToatbCupi0O44diiiLLsZ93Zo=
+go.unistack.org/micro/v3 v3.11.45 h1:fjTLZYWgsVf9FIMZBxOg8ios2/tmyimnjZrsrxEUeXU=
+go.unistack.org/micro/v3 v3.11.45/go.mod h1:fDQ8Mu9wubaFP0L8hNQlpzHiEnWN0wbOlawN9HYo0N4=
+go.unistack.org/micro/v3 v3.11.48 h1:lHJYSHU2z1TTcuswItGwG7cZXN6n04EFqY7lk/0gA7w=
+go.unistack.org/micro/v3 v3.11.48/go.mod h1:fDQ8Mu9wubaFP0L8hNQlpzHiEnWN0wbOlawN9HYo0N4=
+golang.org/x/crypto v0.43.0 h1:dduJYIi3A3KOfdGOHX8AVZ/jGiyPa3IbBozJ5kNuE04=
+golang.org/x/crypto v0.43.0/go.mod h1:BFbav4mRNlXJL4wNeejLpWxB7wMbc79PdRGhWKncxR0=
+golang.org/x/net v0.45.0 h1:RLBg5JKixCy82FtLJpeNlVM0nrSqpCRYzVU1n8kj0tM=
+golang.org/x/net v0.45.0/go.mod h1:ECOoLqd5U3Lhyeyo/QDCEVQ4sNgYsqvCZ722XogGieY=
+golang.org/x/sys v0.37.0 h1:fdNQudmxPjkdUTPnLn5mdQv7Zwvbvpaxqs831goi9kQ=
+golang.org/x/sys v0.37.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
+golang.org/x/text v0.30.0 h1:yznKA/E9zq54KzlzBEAWn1NXSQ8DIp/NYMy88xJjl4k=
+golang.org/x/text v0.30.0/go.mod h1:yDdHFIX9t+tORqspjENWgzaCVXgk0yYnYuSZ8UzzBVM=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20251007200510-49b9836ed3ff h1:A90eA31Wq6HOMIQlLfzFwzqGKBTuaVztYu/g8sn+8Zc=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20251007200510-49b9836ed3ff/go.mod h1:7i2o+ce6H/6BluujYR+kqX3GKH+dChPTQU19wjRPiGk=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20251022142026-3a174f9686a8 h1:M1rk8KBnUsBDg1oPGHNCxG4vc1f49epmTO7xscSajMk=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20251022142026-3a174f9686a8/go.mod h1:7i2o+ce6H/6BluujYR+kqX3GKH+dChPTQU19wjRPiGk=
+google.golang.org/grpc v1.76.0 h1:UnVkv1+uMLYXoIz6o7chp59WfQUYA2ex/BXQ9rHZu7A=
+google.golang.org/grpc v1.76.0/go.mod h1:Ju12QI8M6iQJtbcsV+awF5a4hfJMLi4X0JLo94ULZ6c=
+google.golang.org/protobuf v1.36.10 h1:AYd7cD/uASjIL6Q9LiTjz8JLcrh/88q5UObnmY3aOOE=
+google.golang.org/protobuf v1.36.10/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
-gopkg.in/yaml.v2 v2.2.2/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
-gopkg.in/yaml.v2 v2.3.0/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
-gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
+gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
+gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
+gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
+gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

hook_event.go

@@ -0,0 +1,73 @@
+package kgo
+
+import (
+	"context"
+	"net"
+	"sync/atomic"
+	"time"
+
+	"github.com/twmb/franz-go/pkg/kgo"
+	"go.unistack.org/micro/v3/logger"
+)
+
+type hookEvent struct {
+	log          logger.Logger
+	connected    *atomic.Uint32
+	fatalOnError bool
+}
+
+var (
+	_ kgo.HookBrokerConnect           = &hookEvent{}
+	_ kgo.HookBrokerDisconnect        = &hookEvent{}
+	_ kgo.HookBrokerRead              = &hookEvent{}
+	_ kgo.HookBrokerWrite             = &hookEvent{}
+	_ kgo.HookGroupManageError        = &hookEvent{}
+	_ kgo.HookProduceRecordUnbuffered = &hookEvent{}
+)
+
+func (m *hookEvent) OnGroupManageError(err error) {
+	switch {
+	case err == nil || isContextError(err) || kgo.IsRetryableBrokerErr(err):
+		return
+	default:
+		m.log.Error(context.TODO(), "kgo.OnGroupManageError", err)
+	}
+}
+
+func (m *hookEvent) OnBrokerConnect(_ kgo.BrokerMetadata, _ time.Duration, _ net.Conn, err error) {
+	switch {
+	case err == nil || isContextError(err) || kgo.IsRetryableBrokerErr(err):
+		return
+	default:
+		m.log.Error(context.TODO(), "kgo.OnBrokerConnect", err)
+	}
+}
+
+func (m *hookEvent) OnBrokerDisconnect(_ kgo.BrokerMetadata, _ net.Conn) {}
+
+func (m *hookEvent) OnBrokerWrite(_ kgo.BrokerMetadata, _ int16, _ int, _ time.Duration, _ time.Duration, err error) {
+	switch {
+	case err == nil || isContextError(err) || kgo.IsRetryableBrokerErr(err):
+		return
+	default:
+		m.log.Error(context.TODO(), "kgo.OnBrokerWrite", err)
+	}
+}
+
+func (m *hookEvent) OnBrokerRead(_ kgo.BrokerMetadata, _ int16, _ int, _ time.Duration, _ time.Duration, err error) {
+	switch {
+	case err == nil || isContextError(err) || kgo.IsRetryableBrokerErr(err):
+		return
+	default:
+		m.log.Error(context.TODO(), "kgo.OnBrokerRead", err)
+	}
+}
+
+func (m *hookEvent) OnProduceRecordUnbuffered(_ *kgo.Record, err error) {
+	switch {
+	case err == nil || isContextError(err) || kgo.IsRetryableBrokerErr(err):
+		return
+	default:
+		m.log.Error(context.TODO(), "kgo.OnProduceRecordUnbuffered", err)
+	}
+}

hook_event_test.go

@@ -0,0 +1,362 @@
+package kgo
+
+import (
+	"context"
+	"errors"
+	"io"
+	"net"
+	"os"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+	"github.com/twmb/franz-go/pkg/kgo"
+	"go.unistack.org/micro/v3/logger"
+)
+
+func TestHookEvent_OnGroupManageError(t *testing.T) {
+	tests := []struct {
+		name                  string
+		inputErr              error
+		expectedErrorIsCalled bool
+		expectedErrorMsg      string
+	}{
+		{
+			name:                  "error is nil",
+			inputErr:              nil,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context canceled",
+			inputErr:              context.Canceled,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context deadline exceeded",
+			inputErr:              context.DeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: deadline exceeded (os package)",
+			inputErr:              os.ErrDeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: EOF (io package)",
+			inputErr:              io.EOF,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: closed network connection (net package)",
+			inputErr:              net.ErrClosed,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "some error",
+			inputErr:              errors.New("some error"),
+			expectedErrorIsCalled: true,
+			expectedErrorMsg:      "kgo.OnGroupManageError",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			log := &mockLogger{}
+			he := &hookEvent{log: log}
+			he.OnGroupManageError(tt.inputErr)
+			require.Equal(t, tt.expectedErrorIsCalled, log.errorIsCalled)
+			require.Equal(t, tt.expectedErrorMsg, log.errorMsg)
+		})
+	}
+}
+
+func TestHookEvent_OnBrokerConnect(t *testing.T) {
+	tests := []struct {
+		name                  string
+		inputErr              error
+		expectedErrorIsCalled bool
+		expectedErrorMsg      string
+	}{
+		{
+			name:                  "error is nil",
+			inputErr:              nil,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context canceled",
+			inputErr:              context.Canceled,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context deadline exceeded",
+			inputErr:              context.DeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: deadline exceeded (os package)",
+			inputErr:              os.ErrDeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: EOF (io package)",
+			inputErr:              io.EOF,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: closed network connection (net package)",
+			inputErr:              net.ErrClosed,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "some error",
+			inputErr:              errors.New("some error"),
+			expectedErrorIsCalled: true,
+			expectedErrorMsg:      "kgo.OnBrokerConnect",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			log := &mockLogger{}
+			he := &hookEvent{log: log}
+			he.OnBrokerConnect(kgo.BrokerMetadata{}, 0, nil, tt.inputErr)
+			require.Equal(t, tt.expectedErrorIsCalled, log.errorIsCalled)
+			require.Equal(t, tt.expectedErrorMsg, log.errorMsg)
+		})
+	}
+}
+
+func TestHookEvent_OnBrokerWrite(t *testing.T) {
+	tests := []struct {
+		name                  string
+		inputErr              error
+		expectedErrorIsCalled bool
+		expectedErrorMsg      string
+	}{
+		{
+			name:                  "error is nil",
+			inputErr:              nil,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context canceled",
+			inputErr:              context.Canceled,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context deadline exceeded",
+			inputErr:              context.DeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: deadline exceeded (os package)",
+			inputErr:              os.ErrDeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: EOF (io package)",
+			inputErr:              io.EOF,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: closed network connection (net package)",
+			inputErr:              net.ErrClosed,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "some error",
+			inputErr:              errors.New("some error"),
+			expectedErrorIsCalled: true,
+			expectedErrorMsg:      "kgo.OnBrokerWrite",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			log := &mockLogger{}
+			he := &hookEvent{log: log}
+			he.OnBrokerWrite(kgo.BrokerMetadata{}, 0, 0, 0, 0, tt.inputErr)
+			require.Equal(t, tt.expectedErrorIsCalled, log.errorIsCalled)
+			require.Equal(t, tt.expectedErrorMsg, log.errorMsg)
+		})
+	}
+}
+
+func TestHookEvent_OnBrokerRead(t *testing.T) {
+	tests := []struct {
+		name                  string
+		inputErr              error
+		expectedErrorIsCalled bool
+		expectedErrorMsg      string
+	}{
+		{
+			name:                  "error is nil",
+			inputErr:              nil,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context canceled",
+			inputErr:              context.Canceled,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context deadline exceeded",
+			inputErr:              context.DeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: deadline exceeded (os package)",
+			inputErr:              os.ErrDeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: EOF (io package)",
+			inputErr:              io.EOF,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: closed network connection (net package)",
+			inputErr:              net.ErrClosed,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "some error",
+			inputErr:              errors.New("some error"),
+			expectedErrorIsCalled: true,
+			expectedErrorMsg:      "kgo.OnBrokerRead",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			log := &mockLogger{}
+			he := &hookEvent{log: log}
+			he.OnBrokerRead(kgo.BrokerMetadata{}, 0, 0, 0, 0, tt.inputErr)
+			require.Equal(t, tt.expectedErrorIsCalled, log.errorIsCalled)
+			require.Equal(t, tt.expectedErrorMsg, log.errorMsg)
+		})
+	}
+}
+
+func TestHookEvent_OnProduceRecordUnbuffered(t *testing.T) {
+	tests := []struct {
+		name                  string
+		inputErr              error
+		expectedErrorIsCalled bool
+		expectedErrorMsg      string
+	}{
+		{
+			name:                  "error is nil",
+			inputErr:              nil,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context canceled",
+			inputErr:              context.Canceled,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "context deadline exceeded",
+			inputErr:              context.DeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: deadline exceeded (os package)",
+			inputErr:              os.ErrDeadlineExceeded,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: EOF (io package)",
+			inputErr:              io.EOF,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "retryable error: closed network connection (net package)",
+			inputErr:              net.ErrClosed,
+			expectedErrorIsCalled: false,
+		},
+		{
+			name:                  "some error",
+			inputErr:              errors.New("some error"),
+			expectedErrorIsCalled: true,
+			expectedErrorMsg:      "kgo.OnProduceRecordUnbuffered",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			log := &mockLogger{}
+			he := &hookEvent{log: log}
+			he.OnProduceRecordUnbuffered(&kgo.Record{}, tt.inputErr)
+			require.Equal(t, tt.expectedErrorIsCalled, log.errorIsCalled)
+			require.Equal(t, tt.expectedErrorMsg, log.errorMsg)
+		})
+	}
+}
+
+// Mocks
+
+type mockLogger struct {
+	errorIsCalled bool
+	errorMsg      string
+}
+
+func (m *mockLogger) Init(...logger.Option) error {
+	panic("implement me")
+}
+
+func (m *mockLogger) Clone(...logger.Option) logger.Logger {
+	panic("implement me")
+}
+
+func (m *mockLogger) V(logger.Level) bool {
+	panic("implement me")
+}
+
+func (m *mockLogger) Level(logger.Level) {
+	panic("implement me")
+}
+
+func (m *mockLogger) Options() logger.Options {
+	panic("implement me")
+}
+
+func (m *mockLogger) Fields(...interface{}) logger.Logger {
+	panic("implement me")
+}
+
+func (m *mockLogger) Info(context.Context, string, ...interface{}) {
+	panic("implement me")
+}
+
+func (m *mockLogger) Trace(context.Context, string, ...interface{}) {
+	panic("implement me")
+}
+
+func (m *mockLogger) Debug(context.Context, string, ...interface{}) {
+	panic("implement me")
+}
+
+func (m *mockLogger) Warn(context.Context, string, ...interface{}) {
+	panic("implement me")
+}
+
+func (m *mockLogger) Error(ctx context.Context, msg string, args ...interface{}) {
+	m.errorIsCalled = true
+	m.errorMsg = msg
+}
+
+func (m *mockLogger) Fatal(context.Context, string, ...interface{}) {
+	panic("implement me")
+}
+
+func (m *mockLogger) Log(context.Context, logger.Level, string, ...interface{}) {
+	panic("implement me")
+}
+
+func (m *mockLogger) Name() string {
+	panic("implement me")
+}
+
+func (m *mockLogger) String() string {
+	panic("implement me")
+}

kgo.go

@@ -3,202 +3,305 @@ package kgo
 
 import (
 	"context"
+	"errors"
 	"fmt"
+	"math/rand/v2"
+	"strconv"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
-	"github.com/twmb/franz-go/pkg/kerr"
-	kgo "github.com/twmb/franz-go/pkg/kgo"
+	"github.com/google/uuid"
+	"github.com/twmb/franz-go/pkg/kadm"
+	"github.com/twmb/franz-go/pkg/kgo"
 	"github.com/twmb/franz-go/pkg/kmsg"
-	"github.com/twmb/franz-go/pkg/kversion"
 	"go.unistack.org/micro/v3/broker"
 	"go.unistack.org/micro/v3/logger"
 	"go.unistack.org/micro/v3/metadata"
-	"go.unistack.org/micro/v3/util/id"
+	"go.unistack.org/micro/v3/semconv"
+	"go.unistack.org/micro/v3/tracer"
+	mjitter "go.unistack.org/micro/v3/util/jitter"
 	mrand "go.unistack.org/micro/v3/util/rand"
 )
 
-var _ broker.Broker = &kBroker{}
+var _ broker.Broker = (*Broker)(nil)
 
-type kBroker struct {
-	writer    *kgo.Client // used only to push messages
-	kopts     []kgo.Opt
-	connected bool
-	init      bool
-	sync.RWMutex
-	opts broker.Options
-	subs []*subscriber
+var ErrLostMessage = errors.New("message not marked for offsets commit and will be lost in next iteration")
+
+var DefaultRetryBackoffFn = func() func(int) time.Duration {
+	var rngMu sync.Mutex
+	return func(fails int) time.Duration {
+		const (
+			min = 100 * time.Millisecond
+			max = time.Second
+		)
+		if fails <= 0 {
+			return min
+		}
+		if fails > 10 {
+			return max
 		}
 
-type subscriber struct {
-	reader       *kgo.Client // used only to pull messages
-	topic        string
-	opts         broker.SubscribeOptions
-	kopts        broker.Options
-	handler      broker.Handler
-	batchhandler broker.BatchHandler
-	closed       bool
+		backoff := min * time.Duration(1<<(fails-1))
+
+		rngMu.Lock()
+		jitter := 0.8 + 0.4*rand.Float64()
+		rngMu.Unlock()
+
+		backoff = time.Duration(float64(backoff) * jitter)
+
+		if backoff > max {
+			return max
+		}
+		return backoff
+	}
+}()
+
+type Broker struct {
+	c         *kgo.Client
+	connected *atomic.Uint32
+
 	done chan struct{}
-	consumers    map[string]map[int32]worker
-	sync.RWMutex
+
+	kopts []kgo.Opt
+	subs  []*Subscriber
+
+	opts broker.Options
+	mu   sync.RWMutex
+
+	init bool
 }
 
-type publication struct {
-	topic string
-	err   error
-	sync.RWMutex
-	msg *broker.Message
-	ack bool
+func (r *Broker) Live() bool {
+	return r.connected.Load() == 1
 }
 
-func (p *publication) Topic() string {
-	return p.topic
+func (r *Broker) Ready() bool {
+	return r.connected.Load() == 1
 }
 
-func (p *publication) Message() *broker.Message {
-	return p.msg
+func (r *Broker) Health() bool {
+	return r.connected.Load() == 1
 }
 
-func (p *publication) Ack() error {
-	p.ack = true
-	return nil
-}
-
-func (p *publication) Error() error {
-	return p.err
-}
-
-func (p *publication) SetError(err error) {
-	p.err = err
-}
-
-func (s *subscriber) Options() broker.SubscribeOptions {
-	return s.opts
-}
-
-func (s *subscriber) Topic() string {
-	return s.topic
-}
-
-func (s *subscriber) Unsubscribe(ctx context.Context) error {
-	if s.closed {
-		return nil
-	}
-	select {
-	case <-ctx.Done():
-		return ctx.Err()
-	default:
-		close(s.done)
-		s.closed = true
-	}
-	return nil
-}
-
-func (k *kBroker) Address() string {
+func (k *Broker) Address() string {
 	return strings.Join(k.opts.Addrs, ",")
 }
 
-func (k *kBroker) Name() string {
+func (k *Broker) Name() string {
 	return k.opts.Name
 }
 
-func (k *kBroker) Connect(ctx context.Context) error {
-	k.RLock()
-	if k.connected {
-		k.RUnlock()
+func (k *Broker) Client() *kgo.Client {
+	return k.c
+}
+
+func (k *Broker) connect(ctx context.Context, opts ...kgo.Opt) (*kgo.Client, *hookTracer, error) {
+	var c *kgo.Client
+	var err error
+
+	sp, _ := tracer.SpanFromContext(ctx)
+
+	clientID := "kgo"
+	group := ""
+	if k.opts.Context != nil {
+		if id, ok := k.opts.Context.Value(clientIDKey{}).(string); ok {
+			clientID = id
+		}
+		if id, ok := k.opts.Context.Value(groupKey{}).(string); ok {
+			group = id
+		}
+	}
+
+	var fatalOnError bool
+	if k.opts.Context != nil {
+		if v, ok := k.opts.Context.Value(fatalOnErrorKey{}).(bool); ok && v {
+			fatalOnError = v
+		}
+	}
+
+	htracer := &hookTracer{group: group, clientID: clientID, tracer: k.opts.Tracer}
+	opts = append(opts,
+		kgo.WithHooks(&hookMeter{meter: k.opts.Meter}),
+		kgo.WithHooks(htracer),
+		kgo.WithHooks(&hookEvent{log: k.opts.Logger, fatalOnError: fatalOnError, connected: k.connected}),
+	)
+
+	select {
+	case <-ctx.Done():
+		if ctx.Err() != nil {
+			if sp != nil {
+				sp.SetStatus(tracer.SpanStatusError, ctx.Err().Error())
+			}
+		}
+		return nil, nil, ctx.Err()
+	default:
+		c, err = kgo.NewClient(opts...)
+		if err == nil {
+			err = c.Ping(ctx) // check connectivity to cluster
+		}
+		if err != nil {
+			if sp != nil {
+				sp.SetStatus(tracer.SpanStatusError, err.Error())
+			}
+			return nil, nil, err
+		}
+		k.connected.Store(1)
+
+		if fatalOnError {
+			go func() {
+				c := 3
+				n := 0
+				tc := mjitter.NewTicker(500*time.Millisecond, 1*time.Second)
+				defer tc.Stop()
+				for range tc.C {
+					if k.connected.Load() == 0 {
+						if n > c {
+							k.opts.Logger.Fatal(context.Background(), "broker fatal error")
+						}
+						n++
+					} else {
+						n = 0
+					}
+				}
+			}()
+		}
+		return c, htracer, nil
+	}
+}
+
+func (k *Broker) Connect(ctx context.Context) error {
+	if k.connected.Load() == 1 {
 		return nil
 	}
-	k.RUnlock()
 
 	nctx := k.opts.Context
 	if ctx != nil {
 		nctx = ctx
 	}
 
-	kaddrs := k.opts.Addrs
-
-	// shuffle addrs
-	var rng mrand.Rand
-	rng.Shuffle(len(kaddrs), func(i, j int) {
-		kaddrs[i], kaddrs[j] = kaddrs[j], kaddrs[i]
-	})
-
-	kopts := append(k.kopts, kgo.SeedBrokers(kaddrs...))
-
-	select {
-	case <-nctx.Done():
-		return nctx.Err()
-	default:
-		c, err := kgo.NewClient(kopts...)
+	c, _, err := k.connect(nctx, k.kopts...)
 	if err != nil {
 		return err
 	}
 
-		// Request versions in order to guess Kafka Cluster version
-		versionsReq := kmsg.NewApiVersionsRequest()
-		versionsRes, err := versionsReq.RequestWith(ctx, c)
-		if err != nil {
-			return fmt.Errorf("failed to request api versions: %w", err)
-		}
-		err = kerr.ErrorForCode(versionsRes.ErrorCode)
-		if err != nil {
-			return fmt.Errorf("failed to request api versions. Inner kafka error: %w", err)
-		}
-		versions := kversion.FromApiVersionsResponse(versionsRes)
+	k.mu.Lock()
+	k.c = c
+	k.connected.Store(1)
+	k.mu.Unlock()
 
-		if k.opts.Logger.V(logger.InfoLevel) {
-			logger.Infof(ctx, "[kgo] connected to to kafka cluster version %v", versions.VersionGuess())
+	exposeLag := false
+	if k.opts.Context != nil {
+		if v, ok := k.opts.Context.Value(exposeLagKey{}).(bool); ok && v {
+			exposeLag = v
+		}
+	}
+
+	if exposeLag {
+		var mu sync.Mutex
+		var lastUpdate time.Time
+		type pl struct {
+			p string
+			l float64
+		}
+
+		lag := make(map[string]map[string]pl) // topic => group => partition => lag
+		ac := kadm.NewClient(k.c)
+
+		updateStats := func() {
+			mu.Lock()
+			if time.Since(lastUpdate) < DefaultStatsInterval {
+				return
+			}
+			mu.Unlock()
+
+			k.mu.Lock()
+			groups := make([]string, 0, len(k.subs))
+			for _, g := range k.subs {
+				groups = append(groups, g.opts.Group)
+			}
+			k.mu.Unlock()
+
+			dgls, err := ac.Lag(ctx, groups...)
+			if err != nil || !dgls.Ok() {
+				k.opts.Logger.Error(k.opts.Context, "kgo describe group lag error", err)
+				return
+			}
+
+			for gn, dgl := range dgls {
+				for tn, lmap := range dgl.Lag {
+					if _, ok := lag[tn]; !ok {
+						lag[tn] = make(map[string]pl)
+					}
+					for p, l := range lmap {
+						lag[tn][gn] = pl{p: strconv.Itoa(int(p)), l: float64(l.Lag)}
+					}
+				}
+			}
+		}
+
+		for tn, dg := range lag {
+			for gn, gl := range dg {
+				k.opts.Meter.Gauge(semconv.BrokerGroupLag,
+					func() float64 { updateStats(); return gl.l },
+					"topic", tn,
+					"group", gn,
+					"partition", gl.p)
+			}
 		}
 
-		k.Lock()
-		k.connected = true
-		k.writer = c
-		k.Unlock()
 	}
 
 	return nil
 }
 
-func (k *kBroker) Disconnect(ctx context.Context) error {
-	k.RLock()
-	if !k.connected {
-		k.RUnlock()
+func (k *Broker) Disconnect(ctx context.Context) error {
+	if k.connected.Load() == 0 {
 		return nil
 	}
-	k.RUnlock()
-
-	k.Lock()
-	defer k.Unlock()
 
 	nctx := k.opts.Context
 	if ctx != nil {
 		nctx = ctx
 	}
+	var span tracer.Span
+	ctx, span = k.opts.Tracer.Start(ctx, "Disconnect")
+	defer span.Finish()
 
+	k.mu.Lock()
+	defer k.mu.Unlock()
 	select {
 	case <-nctx.Done():
 		return nctx.Err()
 	default:
 		for _, sub := range k.subs {
+			if sub.closed {
+				continue
+			}
 			if err := sub.Unsubscribe(ctx); err != nil {
 				return err
 			}
 		}
-		k.writer.Close()
+		if k.c != nil {
+			k.c.CloseAllowingRebalance()
+			// k.c.Close()
+		}
 	}
 
-	k.connected = false
+	k.connected.Store(0)
+	close(k.done)
 	return nil
 }
 
-func (k *kBroker) Init(opts ...broker.Option) error {
-	k.Lock()
-	defer k.Unlock()
+func (k *Broker) Init(opts ...broker.Option) error {
+	k.mu.Lock()
+	defer k.mu.Unlock()
 
 	if len(opts) == 0 && k.init {
 		return nil
 	}
+
 	for _, o := range opts {
 		o(&k.opts)
 	}
@@ -222,42 +325,51 @@ func (k *kBroker) Init(opts ...broker.Option) error {
 		}
 	}
 
-	//	kgo.RecordPartitioner(),
-
 	k.init = true
 
 	return nil
 }
 
-func (k *kBroker) Options() broker.Options {
+func (k *Broker) Options() broker.Options {
 	return k.opts
 }
 
-func (k *kBroker) BatchPublish(ctx context.Context, msgs []*broker.Message, opts ...broker.PublishOption) error {
+func (k *Broker) BatchPublish(ctx context.Context, msgs []*broker.Message, opts ...broker.PublishOption) error {
 	return k.publish(ctx, msgs, opts...)
 }
 
-func (k *kBroker) Publish(ctx context.Context, topic string, msg *broker.Message, opts ...broker.PublishOption) error {
+func (k *Broker) Publish(ctx context.Context, topic string, msg *broker.Message, opts ...broker.PublishOption) error {
 	msg.Header.Set(metadata.HeaderTopic, topic)
 	return k.publish(ctx, []*broker.Message{msg}, opts...)
 }
 
-func (k *kBroker) publish(ctx context.Context, msgs []*broker.Message, opts ...broker.PublishOption) error {
+func (k *Broker) publish(ctx context.Context, msgs []*broker.Message, opts ...broker.PublishOption) error {
 	options := broker.NewPublishOptions(opts...)
 	records := make([]*kgo.Record, 0, len(msgs))
 	var errs []string
 	var err error
+	var key []byte
+	var promise func(*kgo.Record, error)
+
+	if options.Context != nil {
+		if k, ok := options.Context.Value(publishKey{}).([]byte); ok && k != nil {
+			key = k
+		}
+		if p, ok := options.Context.Value(publishPromiseKey{}).(func(*kgo.Record, error)); ok && p != nil {
+			promise = p
+		}
+	}
 
 	for _, msg := range msgs {
-		rec := &kgo.Record{}
+		rec := &kgo.Record{Context: ctx, Key: key}
+
 		rec.Topic, _ = msg.Header.Get(metadata.HeaderTopic)
-		if options.BodyOnly {
+		msg.Header.Del(metadata.HeaderTopic)
+
+		k.opts.Meter.Counter(semconv.PublishMessageInflight, "endpoint", rec.Topic, "topic", rec.Topic).Inc()
+		if options.BodyOnly || k.opts.Codec.String() == "noop" {
 			rec.Value = msg.Body
-		} else if k.opts.Codec.String() == "noop" {
-			rec.Value = msg.Body
-			for k, v := range msg.Header {
-				rec.Headers = append(rec.Headers, kgo.RecordHeader{Key: k, Value: []byte(v)})
-			}
+			setHeaders(rec, msg.Header)
 		} else {
 			rec.Value, err = k.opts.Codec.Marshal(msg)
 			if err != nil {
@@ -267,12 +379,41 @@ func (k *kBroker) publish(ctx context.Context, msgs []*broker.Message, opts ...b
 		records = append(records, rec)
 	}
 
-	results := k.writer.ProduceSync(ctx, records...)
+	if promise != nil {
+		ts := time.Now()
+		for _, rec := range records {
+			k.c.Produce(ctx, rec, func(r *kgo.Record, err error) {
+				te := time.Since(ts)
+				k.opts.Meter.Counter(semconv.PublishMessageInflight, "endpoint", rec.Topic, "topic", rec.Topic).Dec()
+				k.opts.Meter.Summary(semconv.PublishMessageLatencyMicroseconds, "endpoint", rec.Topic, "topic", rec.Topic).Update(te.Seconds())
+				k.opts.Meter.Histogram(semconv.PublishMessageDurationSeconds, "endpoint", rec.Topic, "topic", rec.Topic).Update(te.Seconds())
+				if err != nil {
+					k.opts.Meter.Counter(semconv.PublishMessageTotal, "endpoint", rec.Topic, "topic", rec.Topic, "status", "failure").Inc()
+				} else {
+					k.opts.Meter.Counter(semconv.PublishMessageTotal, "endpoint", rec.Topic, "topic", rec.Topic, "status", "success").Inc()
+				}
+				promise(r, err)
+			})
+		}
+		return nil
+	}
+	ts := time.Now()
+
+	results := k.c.ProduceSync(ctx, records...)
+
+	te := time.Since(ts)
 	for _, result := range results {
+		k.opts.Meter.Summary(semconv.PublishMessageLatencyMicroseconds, "endpoint", result.Record.Topic, "topic", result.Record.Topic).Update(te.Seconds())
+		k.opts.Meter.Histogram(semconv.PublishMessageDurationSeconds, "endpoint", result.Record.Topic, "topic", result.Record.Topic).Update(te.Seconds())
+		k.opts.Meter.Counter(semconv.PublishMessageInflight, "endpoint", result.Record.Topic, "topic", result.Record.Topic).Dec()
 		if result.Err != nil {
+			k.opts.Meter.Counter(semconv.PublishMessageTotal, "endpoint", result.Record.Topic, "topic", result.Record.Topic, "status", "failure").Inc()
 			errs = append(errs, result.Err.Error())
+		} else {
+			k.opts.Meter.Counter(semconv.PublishMessageTotal, "endpoint", result.Record.Topic, "topic", result.Record.Topic, "status", "success").Inc()
 		}
 	}
+
 	if len(errs) > 0 {
 		return fmt.Errorf("publish error: %s", strings.Join(errs, "\n"))
 	}
@@ -280,157 +421,134 @@ func (k *kBroker) publish(ctx context.Context, msgs []*broker.Message, opts ...b
 	return nil
 }
 
-type mlogger struct {
-	l   logger.Logger
-	ctx context.Context
+func (k *Broker) TopicExists(ctx context.Context, topic string) error {
+	mdreq := kmsg.NewMetadataRequest()
+	mdreq.Topics = []kmsg.MetadataRequestTopic{
+		{Topic: &topic},
 	}
 
-func (l *mlogger) Log(lvl kgo.LogLevel, msg string, args ...interface{}) {
-	var mlvl logger.Level
-	switch lvl {
-	case kgo.LogLevelNone:
-		return
-	case kgo.LogLevelError:
-		mlvl = logger.ErrorLevel
-	case kgo.LogLevelWarn:
-		mlvl = logger.WarnLevel
-	case kgo.LogLevelInfo:
-		mlvl = logger.InfoLevel
-	case kgo.LogLevelDebug:
-		mlvl = logger.DebugLevel
-	default:
-		return
-	}
-	fields := make(map[string]interface{}, int(len(args)/2))
-	for i := 0; i < len(args)/2; i += 2 {
-		fields[fmt.Sprintf("%v", args[i])] = args[i+1]
-	}
-	l.l.Fields(fields).Log(l.ctx, mlvl, msg)
+	mdrsp, err := mdreq.RequestWith(ctx, k.c)
+	if err != nil {
+		return err
+	} else if mdrsp.Topics[0].ErrorCode != 0 {
+		return fmt.Errorf("topic %s not exists or permission error", topic)
 	}
 
-func (l *mlogger) Level() kgo.LogLevel {
-	switch l.l.Options().Level {
-	case logger.ErrorLevel:
-		return kgo.LogLevelError
-	case logger.WarnLevel:
-		return kgo.LogLevelWarn
-	case logger.InfoLevel:
-		return kgo.LogLevelInfo
-	case logger.DebugLevel, logger.TraceLevel:
-		return kgo.LogLevelDebug
-	}
-	return kgo.LogLevelNone
+	return nil
 }
 
-func (k *kBroker) BatchSubscribe(ctx context.Context, topic string, handler broker.BatchHandler, opts ...broker.SubscribeOption) (broker.Subscriber, error) {
+func (k *Broker) BatchSubscribe(_ context.Context, _ string, _ broker.BatchHandler, _ ...broker.SubscribeOption) (broker.Subscriber, error) {
 	return nil, nil
 }
 
-func (k *kBroker) Subscribe(ctx context.Context, topic string, handler broker.Handler, opts ...broker.SubscribeOption) (broker.Subscriber, error) {
+func (k *Broker) Subscribe(ctx context.Context, topic string, handler broker.Handler, opts ...broker.SubscribeOption) (broker.Subscriber, error) {
 	options := broker.NewSubscribeOptions(opts...)
 
 	if options.Group == "" {
-		uid, err := id.New()
+		uid, err := uuid.NewRandom()
 		if err != nil {
 			return nil, err
 		}
-		options.Group = uid
+		options.Group = uid.String()
 	}
 
-	kaddrs := k.opts.Addrs
+	commitInterval := DefaultCommitInterval
+	if k.opts.Context != nil {
+		if v, ok := k.opts.Context.Value(commitIntervalKey{}).(time.Duration); ok && v > 0 {
+			commitInterval = v
+		}
+	}
 
+	var fatalOnError bool
+	if k.opts.Context != nil {
+		if v, ok := k.opts.Context.Value(fatalOnErrorKey{}).(bool); ok && v {
+			fatalOnError = v
+		}
+	}
+
+	sub := &Subscriber{
+		topic:        topic,
+		opts:         options,
+		handler:      handler,
+		kopts:        k.opts,
+		consumers:    make(map[tp]*consumer),
+		done:         make(chan struct{}),
+		fatalOnError: fatalOnError,
+		connected:    k.connected,
+	}
+
+	kopts := append(k.kopts,
+		kgo.ConsumerGroup(options.Group),
+		kgo.ConsumeTopics(topic),
+		kgo.ConsumeResetOffset(kgo.NewOffset().AtStart()),
+		kgo.FetchMaxWait(1*time.Second),
+		kgo.AutoCommitInterval(commitInterval),
+		kgo.OnPartitionsAssigned(sub.assigned),
+		kgo.OnPartitionsRevoked(sub.revoked),
+		kgo.StopProducerOnDataLossDetected(),
+		kgo.OnPartitionsLost(sub.lost),
+		kgo.AutoCommitCallback(sub.autocommit),
+		kgo.AutoCommitMarks(),
+	)
+
+	if options.Context != nil {
+		if v, ok := options.Context.Value(optionsKey{}).([]kgo.Opt); ok && len(v) > 0 {
+			kopts = append(kopts, v...)
+		}
+	}
+
+	c, htracer, err := k.connect(ctx, kopts...)
+	if err != nil {
+		return nil, err
+	}
+
+	mdreq := kmsg.NewMetadataRequest()
+	mdreq.Topics = []kmsg.MetadataRequestTopic{
+		{Topic: &topic},
+	}
+
+	mdrsp, err := mdreq.RequestWith(ctx, c)
+	if err != nil {
+		return nil, err
+	} else if mdrsp.Topics[0].ErrorCode != 0 {
+		return nil, fmt.Errorf("topic %s not exists or permission error", topic)
+	}
+
+	sub.c = c
+	sub.htracer = htracer
+
+	go sub.poll(ctx)
+
+	k.mu.Lock()
+	k.subs = append(k.subs, sub)
+	k.mu.Unlock()
+	return sub, nil
+}
+
+func (k *Broker) String() string {
+	return "kgo"
+}
+
+func NewBroker(opts ...broker.Option) *Broker {
+	options := broker.NewOptions(opts...)
+
+	kaddrs := options.Addrs
 	// shuffle addrs
 	var rng mrand.Rand
 	rng.Shuffle(len(kaddrs), func(i, j int) {
 		kaddrs[i], kaddrs[j] = kaddrs[j], kaddrs[i]
 	})
-
-	td := DefaultCommitInterval
-	if k.opts.Context != nil {
-		if v, ok := k.opts.Context.Value(commitIntervalKey{}).(time.Duration); ok && v > 0 {
-			td = v
-		}
-	}
-
-	sub := &subscriber{
-		topic:     topic,
-		done:      make(chan struct{}),
-		opts:      options,
-		handler:   handler,
-		kopts:     k.opts,
-		consumers: make(map[string]map[int32]worker),
-	}
-
-	kopts := append(k.kopts,
-		kgo.SeedBrokers(kaddrs...),
-		kgo.ConsumerGroup(options.Group),
-		kgo.ConsumeTopics(topic),
-		kgo.ConsumeResetOffset(kgo.NewOffset().AtStart()),
-		kgo.FetchMaxWait(1*time.Second),
-		// kgo.KeepControlRecords(),
-		kgo.Balancers(kgo.CooperativeStickyBalancer(), kgo.StickyBalancer()),
-		kgo.FetchIsolationLevel(kgo.ReadUncommitted()),
-		kgo.WithHooks(&metrics{meter: k.opts.Meter}),
-		kgo.AutoCommitMarks(),
-		kgo.AutoCommitInterval(td),
-		kgo.OnPartitionsAssigned(sub.assigned),
-		kgo.OnPartitionsRevoked(sub.revoked),
-		kgo.OnPartitionsLost(sub.revoked),
-	)
-
-	reader, err := kgo.NewClient(kopts...)
-	if err != nil {
-		return nil, err
-	}
-
-	sub.reader = reader
-	go sub.run(ctx)
-
-	k.Lock()
-	k.subs = append(k.subs, sub)
-	k.Unlock()
-	return sub, nil
-}
-
-func (k *kBroker) String() string {
-	return "kgo"
-}
-
-func NewBroker(opts ...broker.Option) *kBroker {
-	options := broker.NewOptions(opts...)
-	if options.Codec.String() != "noop" {
-		options.Logger.Infof(options.Context, "broker codec not noop, disable plain kafka headers usage")
-	}
 	kopts := []kgo.Opt{
+		kgo.DialTimeout(3 * time.Second),
 		kgo.DisableIdempotentWrite(),
 		kgo.ProducerBatchCompression(kgo.NoCompression()),
-		kgo.WithLogger(&mlogger{l: options.Logger, ctx: options.Context}),
-		kgo.RetryBackoffFn(
-			func() func(int) time.Duration {
-				var rng mrand.Rand
-				return func(fails int) time.Duration {
-					const (
-						min = 250 * time.Millisecond
-						max = 2 * time.Second
-					)
-					if fails <= 0 {
-						return min
-					}
-					if fails > 10 {
-						return max
-					}
-
-					backoff := min * time.Duration(1<<(fails-1))
-					jitter := 0.8 + 0.4*rng.Float64()
-					backoff = time.Duration(float64(backoff) * jitter)
-
-					if backoff > max {
-						return max
-					}
-					return backoff
-				}
-			}(),
-		),
+		kgo.WithLogger(&mlogger{l: options.Logger.Clone(logger.WithAddCallerSkipCount(2)), ctx: options.Context}),
+		kgo.SeedBrokers(kaddrs...),
+		kgo.RetryBackoffFn(DefaultRetryBackoffFn),
+		kgo.BlockRebalanceOnPoll(),
+		kgo.Balancers(kgo.CooperativeStickyBalancer()),
+		kgo.FetchIsolationLevel(kgo.ReadUncommitted()),
+		kgo.UnknownTopicRetries(1),
 	}
 
 	if options.Context != nil {
@@ -439,8 +557,10 @@ func NewBroker(opts ...broker.Option) *kBroker {
 		}
 	}
 
-	return &kBroker{
+	return &Broker{
+		connected: &atomic.Uint32{},
 		opts:      options,
 		kopts:     kopts,
+		done:      make(chan struct{}),
 	}
 }

kgo_test.go

@@ -2,7 +2,6 @@ package kgo_test
 
 import (
 	"context"
-	"fmt"
 	"os"
 	"strings"
 	"sync/atomic"
@@ -13,14 +12,15 @@ import (
 	kgo "go.unistack.org/micro-broker-kgo/v3"
 	"go.unistack.org/micro/v3/broker"
 	"go.unistack.org/micro/v3/logger"
+	"go.unistack.org/micro/v3/logger/slog"
 	"go.unistack.org/micro/v3/metadata"
 )
 
 var (
-	msgcnt   = int64(12000000)
+	msgcnt   = int64(1200)
 	group    = "38"
-	prefill  = false
-	loglevel = logger.InfoLevel
+	prefill  = true
+	loglevel = logger.DebugLevel
 )
 
 var bm = &broker.Message{
@@ -28,12 +28,121 @@ var bm = &broker.Message{
 	Body:   []byte(`"body"`),
 }
 
+func TestFail(t *testing.T) {
+	if tr := os.Getenv("INTEGRATION_TESTS"); len(tr) > 0 {
+		t.Skip()
+	}
+
+	logger.DefaultLogger = slog.NewLogger()
+	if err := logger.DefaultLogger.Init(logger.WithLevel(loglevel)); err != nil {
+		t.Fatal(err)
+	}
+	ctx := context.Background()
+
+	var addrs []string
+	if addr := os.Getenv("BROKER_ADDRS"); len(addr) == 0 {
+		addrs = []string{"127.0.0.1:9092"}
+	} else {
+		addrs = strings.Split(addr, ",")
+	}
+
+	b := kgo.NewBroker(
+		broker.Addrs(addrs...),
+		kgo.CommitInterval(5*time.Second),
+		kgo.Options(kg.ClientID("test"), kg.FetchMaxBytes(10*1024*1024),
+			kg.AllowAutoTopicCreation(),
+		),
+	)
+
+	t.Logf("broker init")
+	if err := b.Init(); err != nil {
+		t.Fatal(err)
+	}
+
+	t.Logf("broker connect")
+	if err := b.Connect(ctx); err != nil {
+		t.Fatal(err)
+	}
+
+	defer func() {
+		t.Logf("broker disconnect")
+		if err := b.Disconnect(ctx); err != nil {
+			t.Fatal(err)
+		}
+	}()
+
+	t.Logf("broker health %v", b.Health())
+	msgs := make([]*broker.Message, 0, msgcnt)
+	for i := int64(0); i < msgcnt; i++ {
+		msgs = append(msgs, bm)
+	}
+
+	for _, msg := range msgs {
+		t.Logf("broker publish")
+		if err := b.Publish(ctx, "test", msg); err != nil {
+			t.Fatal(err)
+		}
+	}
+	//	t.Skip()
+
+	idx := int64(0)
+	fn := func(msg broker.Event) error {
+		atomic.AddInt64(&idx, 1)
+		time.Sleep(500 * time.Millisecond)
+		t.Logf("ack")
+		return msg.Ack()
+	}
+
+	sub, err := b.Subscribe(ctx, "test", fn,
+		broker.SubscribeAutoAck(true),
+		broker.SubscribeGroup(group),
+		broker.SubscribeBodyOnly(true))
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer func() {
+		if err := sub.Unsubscribe(ctx); err != nil {
+			t.Fatal(err)
+		}
+	}()
+
+	for {
+		t.Logf("health check")
+		if !b.Health() {
+			t.Logf("health works")
+			break
+		}
+		time.Sleep(1 * time.Second)
+	}
+}
+
+func TestConnect(t *testing.T) {
+	if tr := os.Getenv("INTEGRATION_TESTS"); len(tr) > 0 {
+		t.Skip()
+	}
+
+	var addrs []string
+	ctx := context.TODO()
+	b := kgo.NewBroker(
+		broker.Addrs(addrs...),
+		kgo.CommitInterval(5*time.Second),
+		kgo.Options(kg.ClientID("test"), kg.FetchMaxBytes(10*1024*1024)),
+	)
+	if err := b.Init(); err != nil {
+		t.Fatal(err)
+	}
+
+	if err := b.Connect(ctx); err != nil {
+		t.Fatal(err)
+	}
+}
+
 func TestPubSub(t *testing.T) {
 	if tr := os.Getenv("INTEGRATION_TESTS"); len(tr) > 0 {
 		t.Skip()
 	}
 
-	if err := logger.DefaultLogger.Init(logger.WithLevel(loglevel), logger.WithCallerSkipCount(3)); err != nil {
+	if err := logger.DefaultLogger.Init(logger.WithLevel(loglevel)); err != nil {
 		t.Fatal(err)
 	}
 	ctx := context.Background()
@@ -50,6 +159,7 @@ func TestPubSub(t *testing.T) {
 		kgo.CommitInterval(5*time.Second),
 		kgo.Options(kg.ClientID("test"), kg.FetchMaxBytes(10*1024*1024)),
 	)
+
 	if err := b.Init(); err != nil {
 		t.Fatal(err)
 	}
@@ -107,7 +217,7 @@ func TestPubSub(t *testing.T) {
 				if prc := atomic.LoadInt64(&idx); prc == msgcnt {
 					close(done)
 				} else {
-					fmt.Printf("processed %v\n", prc)
+					t.Logf("processed %v\n", prc)
 				}
 			case <-ticker.C:
 				close(done)

logger.go

@@ -0,0 +1,47 @@
+package kgo
+
+import (
+	"context"
+
+	"github.com/twmb/franz-go/pkg/kgo"
+	"go.unistack.org/micro/v3/logger"
+)
+
+type mlogger struct {
+	l   logger.Logger
+	ctx context.Context
+}
+
+func (l *mlogger) Log(lvl kgo.LogLevel, msg string, args ...interface{}) {
+	var mlvl logger.Level
+	switch lvl {
+	case kgo.LogLevelNone:
+		return
+	case kgo.LogLevelError:
+		mlvl = logger.ErrorLevel
+	case kgo.LogLevelWarn:
+		mlvl = logger.WarnLevel
+	case kgo.LogLevelInfo:
+		mlvl = logger.InfoLevel
+	case kgo.LogLevelDebug:
+		mlvl = logger.DebugLevel
+	default:
+		return
+	}
+
+	l.l.Log(l.ctx, mlvl, msg, args...)
+}
+
+func (l *mlogger) Level() kgo.LogLevel {
+	switch l.l.Options().Level {
+	case logger.ErrorLevel:
+		return kgo.LogLevelError
+	case logger.WarnLevel:
+		return kgo.LogLevelWarn
+	case logger.InfoLevel:
+		return kgo.LogLevelInfo
+	case logger.DebugLevel, logger.TraceLevel:
+		return kgo.LogLevelDebug
+	}
+	return kgo.LogLevelNone
+}

meter.go

@@ -0,0 +1,120 @@
+package kgo
+
+import (
+	"net"
+	"strconv"
+	"time"
+
+	"github.com/twmb/franz-go/pkg/kgo"
+	"go.unistack.org/micro/v3/meter"
+)
+
+type hookMeter struct {
+	meter meter.Meter
+}
+
+var (
+	_ kgo.HookBrokerConnect    = &hookMeter{}
+	_ kgo.HookBrokerDisconnect = &hookMeter{}
+	// HookBrokerE2E
+	_ kgo.HookBrokerRead     = &hookMeter{}
+	_ kgo.HookBrokerThrottle = &hookMeter{}
+	_ kgo.HookBrokerWrite    = &hookMeter{}
+	_ kgo.HookFetchBatchRead = &hookMeter{}
+	// HookFetchRecordBuffered
+	// HookFetchRecordUnbuffered
+	_ kgo.HookGroupManageError = &hookMeter{}
+	// HookNewClient
+	_ kgo.HookProduceBatchWritten = &hookMeter{}
+	// HookProduceRecordBuffered
+	// HookProduceRecordPartitioned
+	// HookProduceRecordUnbuffered
+)
+
+const (
+	metricBrokerConnects    = "micro_broker_connects_total"
+	metricBrokerDisconnects = "micro_broker_disconnects_total"
+
+	metricBrokerWriteErrors        = "micro_broker_write_errors_total"
+	metricBrokerWriteBytes         = "micro_broker_write_bytes_total"
+	metricBrokerWriteWaitLatencies = "micro_broker_write_wait_latencies"
+	metricBrokerWriteLatencies     = "micro_broker_write_latencies"
+
+	metricBrokerReadErrors        = "micro_broker_read_errors_total"
+	metricBrokerReadBytes         = "micro_broker_read_bytes_total"
+	metricBrokerReadWaitLatencies = "micro_broker_read_wait_latencies"
+	metricBrokerReadLatencies     = "micro_broker_read_latencies"
+
+	metricBrokerThrottleLatencies = "micro_broker_throttle_latencies"
+
+	metricBrokerProduceBytesCompressed   = "micro_broker_produce_bytes_compressed_total"
+	metricBrokerProduceBytesUncompressed = "micro_broker_produce_bytes_uncompressed_total"
+	metricBrokerFetchBytesCompressed     = "micro_broker_consume_bytes_compressed_total"
+	metricBrokerFetchBytesUncompressed   = "micro_broker_consume_bytes_uncompressed_total"
+
+	metricBrokerGroupErrors = "micro_broker_group_errors_total"
+
+	labelNode    = "node_id"
+	labelSuccess = "success"
+	labelFaulure = "failure"
+	labelStatus  = "status"
+	labelTopic   = "topic"
+)
+
+func (m *hookMeter) OnGroupManageError(_ error) {
+	m.meter.Counter(metricBrokerGroupErrors).Inc()
+}
+
+func (m *hookMeter) OnBrokerConnect(meta kgo.BrokerMetadata, _ time.Duration, _ net.Conn, err error) {
+	node := strconv.Itoa(int(meta.NodeID))
+	if err != nil {
+		m.meter.Counter(metricBrokerConnects, labelNode, node, labelStatus, labelFaulure).Inc()
+		return
+	}
+	m.meter.Counter(metricBrokerConnects, labelNode, node, labelStatus, labelSuccess).Inc()
+}
+
+func (m *hookMeter) OnBrokerDisconnect(meta kgo.BrokerMetadata, _ net.Conn) {
+	node := strconv.Itoa(int(meta.NodeID))
+	m.meter.Counter(metricBrokerDisconnects, labelNode, node).Inc()
+}
+
+func (m *hookMeter) OnBrokerWrite(meta kgo.BrokerMetadata, _ int16, bytesWritten int, writeWait, timeToWrite time.Duration, err error) {
+	node := strconv.Itoa(int(meta.NodeID))
+	if err != nil {
+		m.meter.Counter(metricBrokerWriteErrors, labelNode, node).Inc()
+		return
+	}
+	m.meter.Counter(metricBrokerWriteBytes, labelNode, node).Add(bytesWritten)
+	m.meter.Histogram(metricBrokerWriteWaitLatencies, labelNode, node).Update(writeWait.Seconds())
+	m.meter.Histogram(metricBrokerWriteLatencies, labelNode, node).Update(timeToWrite.Seconds())
+}
+
+func (m *hookMeter) OnBrokerRead(meta kgo.BrokerMetadata, _ int16, bytesRead int, readWait, timeToRead time.Duration, err error) {
+	node := strconv.Itoa(int(meta.NodeID))
+	if err != nil {
+		m.meter.Counter(metricBrokerReadErrors, labelNode, node).Inc()
+		return
+	}
+	m.meter.Counter(metricBrokerReadBytes, labelNode, node).Add(bytesRead)
+
+	m.meter.Histogram(metricBrokerReadWaitLatencies, labelNode, node).Update(readWait.Seconds())
+	m.meter.Histogram(metricBrokerReadLatencies, labelNode, node).Update(timeToRead.Seconds())
+}
+
+func (m *hookMeter) OnBrokerThrottle(meta kgo.BrokerMetadata, throttleInterval time.Duration, _ bool) {
+	node := strconv.Itoa(int(meta.NodeID))
+	m.meter.Histogram(metricBrokerThrottleLatencies, labelNode, node).Update(throttleInterval.Seconds())
+}
+
+func (m *hookMeter) OnProduceBatchWritten(meta kgo.BrokerMetadata, topic string, _ int32, kmetrics kgo.ProduceBatchMetrics) {
+	node := strconv.Itoa(int(meta.NodeID))
+	m.meter.Counter(metricBrokerProduceBytesUncompressed, labelNode, node, labelTopic, topic).Add(kmetrics.UncompressedBytes)
+	m.meter.Counter(metricBrokerProduceBytesCompressed, labelNode, node, labelTopic, topic).Add(kmetrics.CompressedBytes)
+}
+
+func (m *hookMeter) OnFetchBatchRead(meta kgo.BrokerMetadata, topic string, _ int32, kmetrics kgo.FetchBatchMetrics) {
+	node := strconv.Itoa(int(meta.NodeID))
+	m.meter.Counter(metricBrokerFetchBytesUncompressed, labelNode, node, labelTopic, topic).Add(kmetrics.UncompressedBytes)
+	m.meter.Counter(metricBrokerFetchBytesCompressed, labelNode, node, labelTopic, topic).Add(kmetrics.CompressedBytes)
+}

metrics.go

@@ -1,113 +0,0 @@
-package kgo
-
-import (
-	"net"
-	"strconv"
-	"time"
-
-	"github.com/twmb/franz-go/pkg/kgo"
-	"go.unistack.org/micro/v3/meter"
-)
-
-type metrics struct {
-	meter meter.Meter
-}
-
-var (
-	_ kgo.HookBrokerConnect       = &metrics{}
-	_ kgo.HookBrokerDisconnect    = &metrics{}
-	_ kgo.HookBrokerRead          = &metrics{}
-	_ kgo.HookBrokerThrottle      = &metrics{}
-	_ kgo.HookBrokerWrite         = &metrics{}
-	_ kgo.HookFetchBatchRead      = &metrics{}
-	_ kgo.HookProduceBatchWritten = &metrics{}
-	_ kgo.HookGroupManageError    = &metrics{}
-)
-
-const (
-	metricBrokerConnects    = "broker_connects_total"
-	metricBrokerDisconnects = "broker_disconnects_total"
-
-	metricBrokerWriteErrors        = "broker_write_errors_total"
-	metricBrokerWriteBytes         = "broker_write_bytes_total"
-	metricBrokerWriteWaitLatencies = "broker_write_wait_latencies"
-	metricBrokerWriteLatencies     = "broker_write_latencies"
-
-	metricBrokerReadErrors        = "broker_read_errors_total"
-	metricBrokerReadBytes         = "broker_read_bytes_total"
-	metricBrokerReadWaitLatencies = "broker_read_wait_latencies"
-	metricBrokerReadLatencies     = "broker_read_latencies"
-
-	metricBrokerThrottleLatencies = "broker_throttle_latencies"
-
-	metricBrokerProduceBytesCompressed   = "broker_produce_bytes_compressed_total"
-	metricBrokerProduceBytesUncompressed = "broker_produce_bytes_uncompressed_total"
-	metricBrokerFetchBytesCompressed     = "broker_consume_bytes_compressed_total"
-	metricBrokerFetchBytesUncompressed   = "broker_consume_bytes_uncompressed_total"
-
-	metricBrokerGroupErrors = "broker_group_errors_total"
-
-	labelNode    = "node_id"
-	labelSuccess = "success"
-	labelFaulure = "failure"
-	labelStatus  = "status"
-	labelTopic   = "topic"
-)
-
-func (m *metrics) OnGroupManageError(err error) {
-	m.meter.Counter(metricBrokerGroupErrors).Inc()
-}
-
-func (m *metrics) OnBrokerConnect(meta kgo.BrokerMetadata, _ time.Duration, _ net.Conn, err error) {
-	node := strconv.Itoa(int(meta.NodeID))
-	if err != nil {
-		m.meter.Counter(metricBrokerConnects, labelNode, node, labelStatus, labelFaulure).Inc()
-		return
-	}
-	m.meter.Counter(metricBrokerConnects, labelNode, node, labelStatus, labelSuccess).Inc()
-}
-
-func (m *metrics) OnBrokerDisconnect(meta kgo.BrokerMetadata, _ net.Conn) {
-	node := strconv.Itoa(int(meta.NodeID))
-	m.meter.Counter(metricBrokerDisconnects, labelNode, node).Inc()
-}
-
-func (m *metrics) OnBrokerWrite(meta kgo.BrokerMetadata, _ int16, bytesWritten int, writeWait, timeToWrite time.Duration, err error) {
-	node := strconv.Itoa(int(meta.NodeID))
-	if err != nil {
-		m.meter.Counter(metricBrokerWriteErrors, labelNode, node).Inc()
-		return
-	}
-	m.meter.Counter(metricBrokerWriteBytes, labelNode, node).Add(bytesWritten)
-	m.meter.Histogram(metricBrokerWriteWaitLatencies, labelNode, node).Update(writeWait.Seconds())
-	m.meter.Histogram(metricBrokerWriteLatencies, labelNode, node).Update(timeToWrite.Seconds())
-}
-
-func (m *metrics) OnBrokerRead(meta kgo.BrokerMetadata, _ int16, bytesRead int, readWait, timeToRead time.Duration, err error) {
-	node := strconv.Itoa(int(meta.NodeID))
-	if err != nil {
-		m.meter.Counter(metricBrokerReadErrors, labelNode, node).Inc()
-		return
-	}
-	m.meter.Counter(metricBrokerReadBytes, labelNode, node).Add(bytesRead)
-
-	m.meter.Histogram(metricBrokerReadWaitLatencies, labelNode, node).Update(readWait.Seconds())
-	m.meter.Histogram(metricBrokerReadLatencies, labelNode, node).Update(timeToRead.Seconds())
-}
-
-func (m *metrics) OnBrokerThrottle(meta kgo.BrokerMetadata, throttleInterval time.Duration, _ bool) {
-	node := strconv.Itoa(int(meta.NodeID))
-	m.meter.Histogram(metricBrokerThrottleLatencies, labelNode, node).Update(throttleInterval.Seconds())
-}
-
-func (m *metrics) OnProduceBatchWritten(meta kgo.BrokerMetadata, topic string, _ int32, kmetrics kgo.ProduceBatchMetrics) {
-	node := strconv.Itoa(int(meta.NodeID))
-	m.meter.Counter(metricBrokerProduceBytesUncompressed, labelNode, node, labelTopic, topic).Add(kmetrics.UncompressedBytes)
-	m.meter.Counter(metricBrokerProduceBytesCompressed, labelNode, node, labelTopic, topic).Add(kmetrics.CompressedBytes)
-}
-
-func (m *metrics) OnFetchBatchRead(meta kgo.BrokerMetadata, topic string, _ int32, kmetrics kgo.FetchBatchMetrics) {
-	node := strconv.Itoa(int(meta.NodeID))
-	m.meter.Counter(metricBrokerFetchBytesUncompressed, labelNode, node, labelTopic, topic).Add(kmetrics.UncompressedBytes)
-	m.meter.Counter(metricBrokerFetchBytesCompressed, labelNode, node, labelTopic, topic).Add(kmetrics.CompressedBytes)
-}

options.go

@@ -4,13 +4,22 @@ import (
 	"context"
 	"time"
 
-	kgo "github.com/twmb/franz-go/pkg/kgo"
+	"github.com/twmb/franz-go/pkg/kgo"
 	"go.unistack.org/micro/v3/broker"
 	"go.unistack.org/micro/v3/client"
 )
 
+var (
+
 	// DefaultCommitInterval specifies how fast send commit offsets to kafka
-var DefaultCommitInterval = 5 * time.Second
+	DefaultCommitInterval = 5 * time.Second
+
+	// DefaultStatsInterval specifies how fast check consumer lag
+	DefaultStatsInterval = 30 * time.Second
+
+	// DefaultSubscribeMaxInflight specifies how much messages keep inflight
+	DefaultSubscribeMaxInflight = 100
+)
 
 type subscribeContextKey struct{}
 
@@ -48,6 +57,39 @@ func Options(opts ...kgo.Opt) broker.Option {
 	}
 }
 
+// SubscribeOptions pass additional options to broker in Subscribe
+func SubscribeOptions(opts ...kgo.Opt) broker.SubscribeOption {
+	return func(o *broker.SubscribeOptions) {
+		if o.Context == nil {
+			o.Context = context.Background()
+		}
+		options, ok := o.Context.Value(optionsKey{}).([]kgo.Opt)
+		if !ok {
+			options = make([]kgo.Opt, 0, len(opts))
+		}
+		options = append(options, opts...)
+		o.Context = context.WithValue(o.Context, optionsKey{}, options)
+	}
+}
+
+type fatalOnErrorKey struct{}
+
+func FatalOnError(b bool) broker.Option {
+	return broker.SetOption(fatalOnErrorKey{}, b)
+}
+
+type clientIDKey struct{}
+
+func ClientID(id string) broker.Option {
+	return broker.SetOption(clientIDKey{}, id)
+}
+
+type groupKey struct{}
+
+func Group(id string) broker.Option {
+	return broker.SetOption(groupKey{}, id)
+}
+
 type commitIntervalKey struct{}
 
 // CommitInterval specifies interval to send commits
@@ -55,11 +97,33 @@ func CommitInterval(td time.Duration) broker.Option {
 	return broker.SetOption(commitIntervalKey{}, td)
 }
 
-var DefaultSubscribeMaxInflight = 1000
-
 type subscribeMaxInflightKey struct{}
 
-// SubscribeMaxInFlight specifies interval to send commits
+// SubscribeMaxInFlight max queued messages
 func SubscribeMaxInFlight(n int) broker.SubscribeOption {
 	return broker.SetSubscribeOption(subscribeMaxInflightKey{}, n)
 }
+
+// SubscribeMaxInFlight max queued messages
+func SubscribeFatalOnError(b bool) broker.SubscribeOption {
+	return broker.SetSubscribeOption(fatalOnErrorKey{}, b)
+}
+
+type publishPromiseKey struct{}
+
+// PublishPromise set the kafka promise func for Produce
+func PublishPromise(fn func(*kgo.Record, error)) broker.PublishOption {
+	return broker.SetPublishOption(publishPromiseKey{}, fn)
+}
+
+// ClientPublishKey set the kafka message key (client option)
+func ClientPublishPromise(fn func(*kgo.Record, error)) client.PublishOption {
+	return client.SetPublishOption(publishPromiseKey{}, fn)
+}
+
+type exposeLagKey struct{}
+
+// ExposeLag enabled subscriber lag via [meter.Meter]
+func ExposeLag(b bool) broker.Option {
+	return broker.SetOption(exposeLagKey{}, b)
+}

subscriber.go

@@ -0,0 +1,354 @@
+package kgo
+
+import (
+	"context"
+	"fmt"
+	"strconv"
+	"sync"
+	"sync/atomic"
+	"time"
+
+	"github.com/twmb/franz-go/pkg/kgo"
+	"github.com/twmb/franz-go/pkg/kmsg"
+	"go.unistack.org/micro/v3/broker"
+	"go.unistack.org/micro/v3/logger"
+	"go.unistack.org/micro/v3/metadata"
+	"go.unistack.org/micro/v3/semconv"
+	"go.unistack.org/micro/v3/tracer"
+)
+
+type tp struct {
+	t string
+	p int32
+}
+
+type consumer struct {
+	topic string
+
+	c         *kgo.Client
+	htracer   *hookTracer
+	connected *atomic.Uint32
+
+	quit chan struct{}
+	done chan struct{}
+	recs chan kgo.FetchTopicPartition
+
+	handler broker.Handler
+
+	kopts broker.Options
+	opts  broker.SubscribeOptions
+
+	partition int32
+}
+
+type Subscriber struct {
+	topic string
+
+	consumers map[tp]*consumer
+
+	c         *kgo.Client
+	htracer   *hookTracer
+	connected *atomic.Uint32
+
+	handler broker.Handler
+
+	done chan struct{}
+
+	kopts broker.Options
+	opts  broker.SubscribeOptions
+
+	closed       bool
+	fatalOnError bool
+
+	sync.RWMutex
+}
+
+func (s *Subscriber) Client() *kgo.Client {
+	return s.c
+}
+
+func (s *Subscriber) Options() broker.SubscribeOptions {
+	return s.opts
+}
+
+func (s *Subscriber) Topic() string {
+	return s.topic
+}
+
+func (s *Subscriber) Unsubscribe(ctx context.Context) error {
+	if s.closed {
+		return nil
+	}
+
+	s.c.PauseFetchTopics(s.topic)
+	s.c.CloseAllowingRebalance()
+	kc := make(map[string][]int32)
+	for ctp := range s.consumers {
+		kc[ctp.t] = append(kc[ctp.t], ctp.p)
+	}
+	s.killConsumers(ctx, kc)
+	close(s.done)
+	s.closed = true
+	s.c.ResumeFetchTopics(s.topic)
+
+	return nil
+}
+
+func (s *Subscriber) poll(ctx context.Context) {
+	maxInflight := DefaultSubscribeMaxInflight
+
+	if s.opts.Context != nil {
+		if n, ok := s.opts.Context.Value(subscribeMaxInflightKey{}).(int); n > 0 && ok {
+			maxInflight = n
+		}
+	}
+
+	for {
+		select {
+		case <-ctx.Done():
+			s.c.CloseAllowingRebalance()
+			return
+		case <-s.done:
+			return
+		default:
+			fetches := s.c.PollRecords(ctx, maxInflight)
+			if !s.closed && fetches.IsClientClosed() {
+				s.closed = true
+				return
+			}
+			fetches.EachError(func(t string, p int32, err error) {
+				if kgo.IsRetryableBrokerErr(err) {
+					s.kopts.Logger.Error(ctx, fmt.Sprintf("[kgo] fetch topic %s partition %d error", t, p), err)
+				} else {
+					s.kopts.Logger.Fatal(ctx, fmt.Sprintf("[kgo] fetch topic %s partition %d error", t, p), err)
+				}
+			})
+
+			fetches.EachPartition(func(p kgo.FetchTopicPartition) {
+				tps := tp{p.Topic, p.Partition}
+				s.consumers[tps].recs <- p
+			})
+			s.c.AllowRebalance()
+		}
+	}
+}
+
+func (s *Subscriber) killConsumers(ctx context.Context, lost map[string][]int32) {
+	var wg sync.WaitGroup
+	defer wg.Wait()
+
+	for topic, partitions := range lost {
+		for _, partition := range partitions {
+			tps := tp{topic, partition}
+			pc, ok := s.consumers[tps]
+			if !ok {
+				continue
+			}
+			delete(s.consumers, tps)
+			close(pc.quit)
+			if s.kopts.Logger.V(logger.DebugLevel) {
+				s.kopts.Logger.Debug(ctx, fmt.Sprintf("[kgo] waiting for work to finish topic %s partition %d", topic, partition))
+			}
+			wg.Add(1)
+			go func() { <-pc.done; wg.Done() }()
+		}
+	}
+}
+
+func (s *Subscriber) autocommit(_ *kgo.Client, _ *kmsg.OffsetCommitRequest, _ *kmsg.OffsetCommitResponse, err error) {
+	if err != nil {
+		//		s.connected.Store(0)
+		if s.fatalOnError {
+			s.kopts.Logger.Fatal(context.TODO(), "kgo.AutoCommitCallback error", err)
+		}
+	}
+}
+
+func (s *Subscriber) lost(ctx context.Context, _ *kgo.Client, lost map[string][]int32) {
+	if s.kopts.Logger.V(logger.ErrorLevel) {
+		s.kopts.Logger.Error(ctx, fmt.Sprintf("[kgo] lost %#+v", lost))
+	}
+	s.killConsumers(ctx, lost)
+	// s.connected.Store(0)
+}
+
+func (s *Subscriber) revoked(ctx context.Context, c *kgo.Client, revoked map[string][]int32) {
+	if s.kopts.Logger.V(logger.DebugLevel) {
+		s.kopts.Logger.Debug(ctx, fmt.Sprintf("[kgo] revoked %#+v", revoked))
+	}
+	s.killConsumers(ctx, revoked)
+	if err := c.CommitMarkedOffsets(ctx); err != nil {
+		s.kopts.Logger.Error(ctx, "[kgo] revoked CommitMarkedOffsets error", err)
+		// s.connected.Store(0)
+	}
+}
+
+func (s *Subscriber) assigned(_ context.Context, c *kgo.Client, assigned map[string][]int32) {
+	for topic, partitions := range assigned {
+		for _, partition := range partitions {
+			pc := &consumer{
+				c:         c,
+				topic:     topic,
+				partition: partition,
+				htracer:   s.htracer,
+				quit:      make(chan struct{}),
+				done:      make(chan struct{}),
+				recs:      make(chan kgo.FetchTopicPartition, 100),
+				handler:   s.handler,
+				kopts:     s.kopts,
+				opts:      s.opts,
+				connected: s.connected,
+			}
+			s.Lock()
+			s.consumers[tp{topic, partition}] = pc
+			s.Unlock()
+			go pc.consume()
+		}
+	}
+}
+
+func (pc *consumer) consume() {
+	defer close(pc.done)
+	if pc.kopts.Logger.V(logger.DebugLevel) {
+		pc.kopts.Logger.Debug(pc.kopts.Context, fmt.Sprintf("starting, topic %s partition %d", pc.topic, pc.partition))
+		defer pc.kopts.Logger.Debug(pc.kopts.Context, fmt.Sprintf("killing, topic %s partition %d", pc.topic, pc.partition))
+	}
+
+	eh := pc.kopts.ErrorHandler
+	if pc.opts.ErrorHandler != nil {
+		eh = pc.opts.ErrorHandler
+	}
+
+	for {
+		select {
+		case <-pc.quit:
+			return
+		case p := <-pc.recs:
+			for _, record := range p.Records {
+				ctx, sp := pc.htracer.WithProcessSpan(record)
+				ts := time.Now()
+				pc.kopts.Meter.Counter(semconv.SubscribeMessageInflight, "endpoint", record.Topic, "topic", record.Topic).Inc()
+				p := eventPool.Get().(*event)
+				p.msg.Header = nil
+				p.msg.Body = nil
+				p.topic = record.Topic
+				p.err = nil
+				p.ack = false
+				p.msg.Header = metadata.New(len(record.Headers))
+				p.ctx = ctx
+				for _, hdr := range record.Headers {
+					p.msg.Header.Set(hdr.Key, string(hdr.Value))
+				}
+				p.msg.Header.Set("Micro-Offset", strconv.FormatInt(record.Offset, 10))
+				p.msg.Header.Set("Micro-Partition", strconv.FormatInt(int64(record.Partition), 10))
+				p.msg.Header.Set("Micro-Topic", record.Topic)
+				p.msg.Header.Set("Micro-Key", string(record.Key))
+				p.msg.Header.Set("Micro-Timestamp", strconv.FormatInt(record.Timestamp.Unix(), 10))
+				if pc.kopts.Codec.String() == "noop" {
+					p.msg.Body = record.Value
+				} else if pc.opts.BodyOnly {
+					p.msg.Body = record.Value
+				} else {
+					if sp != nil {
+						sp.AddEvent("codec unmarshal start")
+					}
+					err := pc.kopts.Codec.Unmarshal(record.Value, p.msg)
+					if sp != nil {
+						sp.AddEvent("codec unmarshal stop")
+					}
+					if err != nil {
+						if sp != nil {
+							sp.SetStatus(tracer.SpanStatusError, err.Error())
+						}
+						pc.kopts.Meter.Counter(semconv.SubscribeMessageTotal, "endpoint", record.Topic, "topic", record.Topic, "status", "failure").Inc()
+						p.err = err
+						p.msg.Body = record.Value
+						if eh != nil {
+							_ = eh(p)
+							pc.kopts.Meter.Counter(semconv.SubscribeMessageInflight, "endpoint", record.Topic, "topic", record.Topic).Dec()
+							if p.ack {
+								pc.c.MarkCommitRecords(record)
+							} else {
+								eventPool.Put(p)
+								//								pc.connected.Store(0)
+								pc.kopts.Logger.Fatal(pc.kopts.Context, "[kgo] ErrLostMessage wtf?")
+								return
+							}
+							eventPool.Put(p)
+							te := time.Since(ts)
+							pc.kopts.Meter.Summary(semconv.SubscribeMessageLatencyMicroseconds, "endpoint", record.Topic, "topic", record.Topic).Update(te.Seconds())
+							pc.kopts.Meter.Histogram(semconv.SubscribeMessageDurationSeconds, "endpoint", record.Topic, "topic", record.Topic).Update(te.Seconds())
+							continue
+						} else {
+							pc.kopts.Logger.Error(pc.kopts.Context, "[kgo]: unmarshal error", err)
+						}
+						te := time.Since(ts)
+						pc.kopts.Meter.Counter(semconv.SubscribeMessageInflight, "endpoint", record.Topic, "topic", record.Topic).Dec()
+						pc.kopts.Meter.Summary(semconv.SubscribeMessageLatencyMicroseconds, "endpoint", record.Topic, "topic", record.Topic).Update(te.Seconds())
+						pc.kopts.Meter.Histogram(semconv.SubscribeMessageDurationSeconds, "endpoint", record.Topic, "topic", record.Topic).Update(te.Seconds())
+						eventPool.Put(p)
+						//						pc.connected.Store(0)
+						pc.kopts.Logger.Fatal(pc.kopts.Context, "[kgo] Unmarshal err not handled wtf?")
+						if sp != nil {
+							sp.Finish()
+						}
+						return
+					}
+				}
+				if sp != nil {
+					sp.AddEvent("handler start")
+				}
+				err := pc.handler(p)
+				if sp != nil {
+					sp.AddEvent("handler stop")
+				}
+				if err == nil {
+					pc.kopts.Meter.Counter(semconv.SubscribeMessageTotal, "endpoint", record.Topic, "topic", record.Topic, "status", "success").Inc()
+				} else {
+					if sp != nil {
+						sp.SetStatus(tracer.SpanStatusError, err.Error())
+					}
+					pc.kopts.Meter.Counter(semconv.SubscribeMessageTotal, "endpoint", record.Topic, "topic", record.Topic, "status", "failure").Inc()
+				}
+				pc.kopts.Meter.Counter(semconv.SubscribeMessageInflight, "endpoint", record.Topic, "topic", record.Topic).Dec()
+				if err == nil && pc.opts.AutoAck {
+					p.ack = true
+				} else if err != nil {
+					p.err = err
+					if eh != nil {
+						if sp != nil {
+							sp.AddEvent("error handler start")
+						}
+						_ = eh(p)
+						if sp != nil {
+							sp.AddEvent("error handler stop")
+						}
+					} else {
+						if pc.kopts.Logger.V(logger.ErrorLevel) {
+							pc.kopts.Logger.Error(pc.kopts.Context, "[kgo]: subscriber error", err)
+						}
+					}
+				}
+				te := time.Since(ts)
+				pc.kopts.Meter.Summary(semconv.SubscribeMessageLatencyMicroseconds, "endpoint", record.Topic, "topic", record.Topic).Update(te.Seconds())
+				pc.kopts.Meter.Histogram(semconv.SubscribeMessageDurationSeconds, "endpoint", record.Topic, "topic", record.Topic).Update(te.Seconds())
+				if p.ack {
+					eventPool.Put(p)
+					pc.c.MarkCommitRecords(record)
+				} else {
+					eventPool.Put(p)
+					//					pc.connected.Store(0)
+					pc.kopts.Logger.Fatal(pc.kopts.Context, "[kgo] ErrLostMessage wtf?")
+					if sp != nil {
+						sp.SetStatus(tracer.SpanStatusError, "ErrLostMessage")
+						sp.Finish()
+					}
+					return
+				}
+				if sp != nil {
+					sp.Finish()
+				}
+			}
+		}
+	}
+}

tracer.go

@@ -0,0 +1,223 @@
+package kgo
+
+import (
+	"context"
+	"unicode/utf8"
+
+	"github.com/twmb/franz-go/pkg/kgo"
+	semconv "go.opentelemetry.io/otel/semconv/v1.18.0"
+	"go.unistack.org/micro/v3/metadata"
+	"go.unistack.org/micro/v3/tracer"
+)
+
+type hookTracer struct {
+	tracer   tracer.Tracer
+	clientID string
+	group    string
+}
+
+var messagingSystem = semconv.MessagingSystemKey.String("kafka")
+
+var (
+	_ kgo.HookProduceRecordBuffered   = (*hookTracer)(nil)
+	_ kgo.HookProduceRecordUnbuffered = (*hookTracer)(nil)
+	_ kgo.HookFetchRecordBuffered     = (*hookTracer)(nil)
+	_ kgo.HookFetchRecordUnbuffered   = (*hookTracer)(nil)
+)
+
+// OnProduceRecordBuffered starts a new span for the "publish" operation on a
+// buffered record.
+//
+// It sets span options and injects the span context into record and updates
+// the record's context, so it can be ended in the OnProduceRecordUnbuffered
+// hook.
+func (m *hookTracer) OnProduceRecordBuffered(r *kgo.Record) {
+	if !m.tracer.Enabled() {
+		return
+	}
+	// Set up span options.
+	attrs := []interface{}{
+		messagingSystem,
+		semconv.MessagingDestinationKindTopic,
+		semconv.MessagingDestinationName(r.Topic),
+		semconv.MessagingOperationPublish,
+	}
+	attrs = maybeKeyAttr(attrs, r)
+	if m.clientID != "" {
+		attrs = append(attrs, semconv.MessagingKafkaClientIDKey.String(m.clientID))
+	}
+	opts := []tracer.SpanOption{
+		tracer.WithSpanLabels(attrs...),
+		tracer.WithSpanKind(tracer.SpanKindProducer),
+	}
+
+	if r.Context == nil {
+		r.Context = context.Background()
+	}
+
+	omd, ok := metadata.FromOutgoingContext(r.Context)
+	if !ok {
+		omd = metadata.New(len(r.Headers))
+	}
+
+	md := metadata.Copy(omd)
+	for _, h := range r.Headers {
+		md.Set(h.Key, string(h.Value))
+	}
+
+	if !ok {
+		r.Context, _ = m.tracer.Start(metadata.NewOutgoingContext(r.Context, md), "sdk.broker", opts...)
+	} else {
+		r.Context, _ = m.tracer.Start(r.Context, "sdk.broker", opts...)
+	}
+
+	setHeaders(r, omd, metadata.HeaderContentType)
+}
+
+// OnProduceRecordUnbuffered continues and ends the "publish" span for an
+// unbuffered record.
+//
+// It sets attributes with values unset when producing and records any error
+// that occurred during the publish operation.
+func (m *hookTracer) OnProduceRecordUnbuffered(r *kgo.Record, err error) {
+	if !m.tracer.Enabled() {
+		return
+	}
+	if span, ok := tracer.SpanFromContext(r.Context); ok {
+		span.AddLabels(
+			semconv.MessagingKafkaDestinationPartition(int(r.Partition)),
+		)
+		if err != nil {
+			span.SetStatus(tracer.SpanStatusError, err.Error())
+		}
+		span.Finish()
+	}
+}
+
+// OnFetchRecordBuffered starts a new span for the "receive" operation on a
+// buffered record.
+//
+// It sets the span options and extracts the span context from the record,
+// updates the record's context to ensure it can be ended in the
+// OnFetchRecordUnbuffered hook and can be used in downstream consumer
+// processing.
+func (m *hookTracer) OnFetchRecordBuffered(r *kgo.Record) {
+	if !m.tracer.Enabled() {
+		return
+	}
+	// Set up the span options.
+	attrs := []interface{}{
+		messagingSystem,
+		semconv.MessagingSourceKindTopic,
+		semconv.MessagingSourceName(r.Topic),
+		semconv.MessagingOperationReceive,
+		semconv.MessagingKafkaSourcePartition(int(r.Partition)),
+	}
+	attrs = maybeKeyAttr(attrs, r)
+	if m.clientID != "" {
+		attrs = append(attrs, semconv.MessagingKafkaClientIDKey.String(m.clientID))
+	}
+	if m.group != "" {
+		attrs = append(attrs, semconv.MessagingKafkaConsumerGroupKey.String(m.group))
+	}
+	opts := []tracer.SpanOption{
+		tracer.WithSpanLabels(attrs...),
+		tracer.WithSpanKind(tracer.SpanKindConsumer),
+	}
+
+	if r.Context == nil {
+		r.Context = context.Background()
+	}
+	omd, ok := metadata.FromIncomingContext(r.Context)
+	if !ok {
+		omd = metadata.New(len(r.Headers))
+	}
+
+	md := metadata.Copy(omd)
+	for _, h := range r.Headers {
+		md.Set(h.Key, string(h.Value))
+	}
+
+	if !ok {
+		r.Context, _ = m.tracer.Start(metadata.NewIncomingContext(r.Context, md), "sdk.broker", opts...)
+	} else {
+		r.Context, _ = m.tracer.Start(r.Context, "sdk.broker", opts...)
+	}
+
+	setHeaders(r, omd, metadata.HeaderContentType)
+}
+
+// OnFetchRecordUnbuffered continues and ends the "receive" span for an
+// unbuffered record.
+func (m *hookTracer) OnFetchRecordUnbuffered(r *kgo.Record, _ bool) {
+	if !m.tracer.Enabled() {
+		return
+	}
+	span, _ := tracer.SpanFromContext(r.Context)
+	span.Finish()
+}
+
+// WithProcessSpan starts a new span for the "process" operation on a consumer
+// record.
+//
+// It sets up the span options. The user's application code is responsible for
+// ending the span.
+//
+// This should only ever be called within a polling loop of a consumed record and
+// not a record which has been created for producing, so call this at the start of each
+// iteration of your processing for the record.
+func (m *hookTracer) WithProcessSpan(r *kgo.Record) (context.Context, tracer.Span) {
+	if r.Context == nil {
+		r.Context = context.Background()
+	}
+
+	if !m.tracer.Enabled() {
+		return r.Context, nil
+	}
+	// Set up the span options.
+	attrs := []interface{}{
+		messagingSystem,
+		semconv.MessagingSourceKindTopic,
+		semconv.MessagingSourceName(r.Topic),
+		semconv.MessagingOperationProcess,
+		semconv.MessagingKafkaSourcePartition(int(r.Partition)),
+		semconv.MessagingKafkaMessageOffset(int(r.Offset)),
+	}
+	attrs = maybeKeyAttr(attrs, r)
+	if m.clientID != "" {
+		attrs = append(attrs, semconv.MessagingKafkaClientIDKey.String(m.clientID))
+	}
+	if m.group != "" {
+		attrs = append(attrs, semconv.MessagingKafkaConsumerGroupKey.String(m.group))
+	}
+	opts := []tracer.SpanOption{
+		tracer.WithSpanLabels(attrs...),
+		tracer.WithSpanKind(tracer.SpanKindConsumer),
+	}
+
+	if r.Context == nil {
+		r.Context = context.Background()
+	}
+	md, ok := metadata.FromIncomingContext(r.Context)
+	if !ok {
+		md = metadata.New(len(r.Headers))
+	}
+	for _, h := range r.Headers {
+		md.Set(h.Key, string(h.Value))
+	}
+
+	// Start a new span using the provided context and options.
+	return m.tracer.Start(r.Context, "sdk.broker", opts...)
+}
+
+func maybeKeyAttr(attrs []interface{}, r *kgo.Record) []interface{} {
+	if r.Key == nil {
+		return attrs
+	}
+	var keykey string
+	if !utf8.Valid(r.Key) {
+		return attrs
+	}
+	keykey = string(r.Key)
+	return append(attrs, semconv.MessagingKafkaMessageKeyKey.String(keykey))
+}

util.go

@@ -1,220 +0,0 @@
-package kgo
-
-import (
-	"context"
-	"errors"
-	"sync"
-
-	kgo "github.com/twmb/franz-go/pkg/kgo"
-	"go.unistack.org/micro/v3/broker"
-	"go.unistack.org/micro/v3/logger"
-	"go.unistack.org/micro/v3/metadata"
-)
-
-var ErrLostMessage = errors.New("message not marked for offsets commit and will be lost in next iteration")
-
-var pPool = sync.Pool{
-	New: func() interface{} {
-		return &publication{msg: broker.NewMessage("")}
-	},
-}
-
-type worker struct {
-	done         chan struct{}
-	recs         chan []*kgo.Record
-	cherr        chan error
-	handler      broker.Handler
-	batchHandler broker.BatchHandler
-	opts         broker.SubscribeOptions
-	kopts        broker.Options
-	tpmap        map[string][]int32
-	maxInflight  int
-	reader       *kgo.Client
-	ctx          context.Context
-}
-
-func (s *subscriber) run(ctx context.Context) {
-	for {
-		select {
-		case <-ctx.Done():
-			return
-		case <-s.kopts.Context.Done():
-			return
-		default:
-			fetches := s.reader.PollFetches(ctx)
-			if fetches.IsClientClosed() {
-				// TODO: fatal ?
-				return
-			}
-			if len(fetches.Errors()) > 0 {
-				for _, err := range fetches.Errors() {
-					s.kopts.Logger.Fatalf(ctx, "fetch err topic %s partition %d: %v", err.Topic, err.Partition, err.Err)
-				}
-				// TODO: fatal ?
-				return
-			}
-
-			fetches.EachPartition(func(p kgo.FetchTopicPartition) {
-				s.Lock()
-				consumers := s.consumers[p.Topic]
-				s.Unlock()
-				if consumers == nil {
-					return
-				}
-				w, ok := consumers[p.Partition]
-				if !ok {
-					return
-				}
-				select {
-				case err := <-w.cherr:
-					s.kopts.Logger.Fatalf(ctx, "handle err: %v", err)
-					return
-				case w.recs <- p.Records:
-				case <-w.done:
-				}
-			})
-		}
-	}
-}
-
-func (s *subscriber) assigned(ctx context.Context, _ *kgo.Client, assigned map[string][]int32) {
-	maxInflight := DefaultSubscribeMaxInflight
-
-	if s.opts.Context != nil {
-		if n, ok := s.opts.Context.Value(subscribeMaxInflightKey{}).(int); n > 0 && ok {
-			maxInflight = n
-		}
-	}
-
-	s.Lock()
-	for topic, partitions := range assigned {
-		if s.consumers[topic] == nil {
-			s.consumers[topic] = make(map[int32]worker)
-		}
-		for _, partition := range partitions {
-			w := worker{
-				done:         make(chan struct{}),
-				recs:         make(chan []*kgo.Record),
-				cherr:        make(chan error),
-				kopts:        s.kopts,
-				opts:         s.opts,
-				ctx:          ctx,
-				tpmap:        map[string][]int32{topic: []int32{partition}},
-				reader:       s.reader,
-				handler:      s.handler,
-				batchHandler: s.batchhandler,
-				maxInflight:  maxInflight,
-			}
-			s.consumers[topic][partition] = w
-			go w.handle()
-		}
-	}
-	s.Unlock()
-}
-
-func (s *subscriber) revoked(_ context.Context, _ *kgo.Client, revoked map[string][]int32) {
-	s.Lock()
-	for topic, partitions := range revoked {
-		ptopics := s.consumers[topic]
-		for _, partition := range partitions {
-			w := ptopics[partition]
-			delete(ptopics, partition)
-			if len(ptopics) == 0 {
-				delete(s.consumers, topic)
-			}
-			close(w.done)
-		}
-	}
-	s.Unlock()
-}
-
-func (w *worker) handle() {
-	var err error
-
-	eh := w.kopts.ErrorHandler
-	if w.opts.ErrorHandler != nil {
-		eh = w.opts.ErrorHandler
-	}
-
-	paused := false
-	for {
-		select {
-		case <-w.ctx.Done():
-			w.cherr <- w.ctx.Err()
-			return
-		case <-w.done:
-			return
-		case recs := <-w.recs:
-			if len(recs) >= w.maxInflight {
-				paused = true
-				w.reader.PauseFetchPartitions(w.tpmap)
-			}
-			for _, record := range recs {
-				p := pPool.Get().(*publication)
-				p.msg.Header = nil
-				p.msg.Body = nil
-				p.topic = record.Topic
-				p.err = nil
-				p.ack = false
-				if w.opts.BodyOnly {
-					p.msg.Body = record.Value
-				} else if w.kopts.Codec.String() == "noop" {
-					p.msg.Body = record.Value
-					p.msg.Header = metadata.New(len(record.Headers))
-					for _, h := range record.Headers {
-						p.msg.Header.Set(h.Key, string(h.Value))
-					}
-				} else {
-					if err := w.kopts.Codec.Unmarshal(record.Value, p.msg); err != nil {
-						p.err = err
-						p.msg.Body = record.Value
-						if eh != nil {
-							_ = eh(p)
-							if p.ack {
-								w.reader.MarkCommitRecords(record)
-							} else {
-								w.cherr <- ErrLostMessage
-								pPool.Put(p)
-								return
-							}
-							pPool.Put(p)
-							continue
-						} else {
-							if w.kopts.Logger.V(logger.ErrorLevel) {
-								w.kopts.Logger.Errorf(w.kopts.Context, "[kgo]: failed to unmarshal: %v", err)
-							}
-						}
-						pPool.Put(p)
-						w.cherr <- err
-						return
-					}
-				}
-				err = w.handler(p)
-				if err == nil && w.opts.AutoAck {
-					p.ack = true
-				} else if err != nil {
-					p.err = err
-					if eh != nil {
-						_ = eh(p)
-					} else {
-						if w.kopts.Logger.V(logger.ErrorLevel) {
-							w.kopts.Logger.Errorf(w.kopts.Context, "[kgo]: subscriber error: %v", err)
-						}
-					}
-				}
-				if p.ack {
-					pPool.Put(p)
-					w.reader.MarkCommitRecords(record)
-				} else {
-					pPool.Put(p)
-					w.cherr <- ErrLostMessage
-					return
-				}
-			}
-			if paused {
-				paused = false
-				w.reader.ResumeFetchPartitions(w.tpmap)
-			}
-		}
-	}
-}