阶段2(上):后端专题 Go/Gin/PostgreSQL/MySQL/Redis/MQ/gRPC 的课程、企业场景、面试题和可运行示例
This commit is contained in:
1 parent
0bae9916a9
commit
5eb67c5d08
65 files changed
+6457
-77
No files matched your search
@@ -0,0 +1,3 @@
|
||||
module example/resilience
|
||||
|
||||
go 1.22
|
||||
@@ -0,0 +1,119 @@
|
||||
// Package resilience 服务调用的三件套:超时(由调用方的 context 控制)、带退避的重试、熔断器。只用标准库。
|
||||
package resilience
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"math/rand/v2"
|
||||
"sync"
|
||||
"time"
|
||||
)
|
||||
|
||||
// ErrOpen 熔断器打开时直接返回,不再调用下游
|
||||
var ErrOpen = errors.New("熔断中:下游暂时不可用")
|
||||
|
||||
// Permanent 包装不应该重试的错误(参数错误、没有权限、资源不存在等)
|
||||
type Permanent struct{ Err error }
|
||||
|
||||
func (p Permanent) Error() string { return p.Err.Error() }
|
||||
func (p Permanent) Unwrap() error { return p.Err }
|
||||
|
||||
// Retry 最多尝试 attempts 次,间隔指数增长并加随机抖动;context 取消或超时立即停止
|
||||
func Retry(ctx context.Context, attempts int, base time.Duration, fn func(context.Context) error) error {
|
||||
var err error
|
||||
for i := range attempts {
|
||||
if err = fn(ctx); err == nil {
|
||||
return nil
|
||||
}
|
||||
var p Permanent
|
||||
if errors.As(err, &p) || errors.Is(err, ErrOpen) || i == attempts-1 {
|
||||
return err
|
||||
}
|
||||
backoff := base << i // base, 2*base, 4*base ...
|
||||
backoff += time.Duration(rand.Int64N(int64(backoff)/2 + 1)) // 抖动,避免所有客户端同时重试
|
||||
select {
|
||||
case <-ctx.Done():
|
||||
return errors.Join(err, ctx.Err())
|
||||
case <-time.After(backoff):
|
||||
}
|
||||
}
|
||||
return err
|
||||
}
|
||||
|
||||
type state int
|
||||
|
||||
const (
|
||||
closed state = iota // 正常放行
|
||||
open // 熔断:直接拒绝
|
||||
halfOpen // 试探:放一个请求过去
|
||||
)
|
||||
|
||||
// Breaker 连续失败 threshold 次后打开,cooldown 之后进入半开,试探成功则关闭、失败则重新打开
|
||||
type Breaker struct {
|
||||
mu sync.Mutex
|
||||
state state
|
||||
failures int
|
||||
threshold int
|
||||
cooldown time.Duration
|
||||
openedAt time.Time
|
||||
probing bool
|
||||
now func() time.Time // 测试时可以替换
|
||||
}
|
||||
|
||||
func NewBreaker(threshold int, cooldown time.Duration) *Breaker {
|
||||
return &Breaker{threshold: threshold, cooldown: cooldown, now: time.Now}
|
||||
}
|
||||
|
||||
func (b *Breaker) allow() bool {
|
||||
b.mu.Lock()
|
||||
defer b.mu.Unlock()
|
||||
switch b.state {
|
||||
case open:
|
||||
if b.now().Sub(b.openedAt) < b.cooldown {
|
||||
return false
|
||||
}
|
||||
b.state = halfOpen
|
||||
fallthrough
|
||||
case halfOpen:
|
||||
if b.probing {
|
||||
return false // 半开状态只放一个试探请求
|
||||
}
|
||||
b.probing = true
|
||||
return true
|
||||
}
|
||||
return true
|
||||
}
|
||||
|
||||
func (b *Breaker) record(err error) {
|
||||
b.mu.Lock()
|
||||
defer b.mu.Unlock()
|
||||
var p Permanent
|
||||
failed := err != nil && !errors.As(err, &p) // 业务错误不算下游故障
|
||||
if b.state == halfOpen {
|
||||
b.probing = false
|
||||
if failed {
|
||||
b.state, b.openedAt = open, b.now()
|
||||
} else {
|
||||
b.state, b.failures = closed, 0
|
||||
}
|
||||
return
|
||||
}
|
||||
if !failed {
|
||||
b.failures = 0
|
||||
return
|
||||
}
|
||||
b.failures++
|
||||
if b.failures >= b.threshold {
|
||||
b.state, b.openedAt = open, b.now()
|
||||
}
|
||||
}
|
||||
|
||||
// Do 通过熔断器调用 fn
|
||||
func (b *Breaker) Do(ctx context.Context, fn func(context.Context) error) error {
|
||||
if !b.allow() {
|
||||
return ErrOpen
|
||||
}
|
||||
err := fn(ctx)
|
||||
b.record(err)
|
||||
return err
|
||||
}
|
||||
@@ -0,0 +1,82 @@
|
||||
package resilience
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"sync/atomic"
|
||||
"testing"
|
||||
"time"
|
||||
)
|
||||
|
||||
var errDown = errors.New("下游 503")
|
||||
|
||||
func TestRetrySucceedsAfterTransientErrors(t *testing.T) {
|
||||
var calls atomic.Int32
|
||||
err := Retry(context.Background(), 5, time.Millisecond, func(context.Context) error {
|
||||
if calls.Add(1) < 3 {
|
||||
return errDown
|
||||
}
|
||||
return nil
|
||||
})
|
||||
if err != nil || calls.Load() != 3 {
|
||||
t.Fatalf("err=%v calls=%d", err, calls.Load())
|
||||
}
|
||||
}
|
||||
|
||||
func TestRetryStopsOnPermanent(t *testing.T) {
|
||||
var calls int
|
||||
err := Retry(context.Background(), 5, time.Millisecond, func(context.Context) error {
|
||||
calls++
|
||||
return Permanent{errors.New("参数错误")}
|
||||
})
|
||||
if calls != 1 || err == nil {
|
||||
t.Fatalf("不可重试的错误只应调用一次,calls=%d", calls)
|
||||
}
|
||||
}
|
||||
|
||||
func TestRetryRespectsDeadline(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Millisecond)
|
||||
defer cancel()
|
||||
start := time.Now()
|
||||
err := Retry(ctx, 10, 20*time.Millisecond, func(context.Context) error { return errDown })
|
||||
if !errors.Is(err, context.DeadlineExceeded) || time.Since(start) > 200*time.Millisecond {
|
||||
t.Fatalf("应该在截止时间后停止:%v,耗时 %v", err, time.Since(start))
|
||||
}
|
||||
}
|
||||
|
||||
func TestBreaker(t *testing.T) {
|
||||
now := time.Unix(0, 0)
|
||||
b := NewBreaker(3, 10*time.Second)
|
||||
b.now = func() time.Time { return now }
|
||||
ctx := context.Background()
|
||||
fail := func(context.Context) error { return errDown }
|
||||
ok := func(context.Context) error { return nil }
|
||||
|
||||
for range 3 {
|
||||
_ = b.Do(ctx, fail)
|
||||
}
|
||||
if err := b.Do(ctx, ok); !errors.Is(err, ErrOpen) {
|
||||
t.Fatalf("连续失败 3 次后应该熔断,得到 %v", err)
|
||||
}
|
||||
now = now.Add(11 * time.Second) // 冷却结束 → 半开,试探失败 → 重新打开
|
||||
if err := b.Do(ctx, fail); !errors.Is(err, errDown) {
|
||||
t.Fatalf("半开时应放行试探请求:%v", err)
|
||||
}
|
||||
if err := b.Do(ctx, ok); !errors.Is(err, ErrOpen) {
|
||||
t.Fatal("试探失败后应该重新熔断")
|
||||
}
|
||||
now = now.Add(11 * time.Second) // 再次半开,试探成功 → 关闭
|
||||
if err := b.Do(ctx, ok); err != nil {
|
||||
t.Fatal(err)
|
||||
}
|
||||
if err := b.Do(ctx, ok); err != nil {
|
||||
t.Fatal("恢复后应该正常放行")
|
||||
}
|
||||
// 业务错误不计入熔断
|
||||
for range 5 {
|
||||
_ = b.Do(ctx, func(context.Context) error { return Permanent{errors.New("用户不存在")} })
|
||||
}
|
||||
if err := b.Do(ctx, ok); err != nil {
|
||||
t.Fatal("业务错误不应该触发熔断")
|
||||
}
|
||||
}
|
||||
Reference in new issue
Block a user