使用Redis实现消息队列

07 Jul 2021

10 minutes reading time

使用Redis实现消息队列主要有三种方法:

下面分别对这三种方法进行介绍,并编写简单例子。

#List队列

看到队列,你会想到Redis有个数据类型List,List能很好符合队列的要求。List的底层是一个链表,在头部和尾部进行操作的时间复杂度都是O(1)。 使用List进行队列操作,你可以这样使用。 生产者使用LPUSH进行消息发布 image.png 消费者使用RPOP对消息进行消费 image.png 这里存在着一个问题,如果LIST没有消息时,消费者执行RPOP时,会返回null(nil)。 我们在编写消费者逻辑时,一般是循环不断从队列中消费数据进行处理,如果此时队列为空,那消费者依旧会频繁拉取消息,这会造成「CPU 空转」,不仅浪费 CPU 资源,还会对 Redis 造成压力。Redis提供了阻塞式拉起命令BRPOP / BLPOP,使用 BRPOP 这种阻塞式方式拉取消息时,还支持传入一个「超时时间」,如果设置为 0,则表示不设置超时,直到有新消息才返回,否则会在指定的超时时间后返回 null,既兼顾了效率还避免了CPU空转问题。 这是List队列的代码例子:

// 生产者
package main

import (
	"context"
    
	"github.com/go-redis/redis/v8"
)

func main() {
	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "",
		DB:       1,
	})
	data :=[]string{`{"name": "jinzhu11221212321", "age": 18, "tags": ["tag211", "tag2"], "orgs": {"orga": "orga"}`,`{"name": "asd", "age": 18, "tags": ["tag211", "tag2"], "orgs": {"orga": "orga"}`}
	rdb.LPush(context.Background(), "queue", data)
}

// 消费者
package main

import (
	"context"
	"fmt"
	"log"
	"time"

	"github.com/go-redis/redis/v8"
)

func main() {
	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "",
		DB:       1,
	})
	for true {
		//result, err := rdb.RPop(context.Background(), "queue").Result()
		result, err := rdb.BRPop(context.Background(), 0, "queue").Result()
		if err == redis.Nil {
			continue
		} else if err != nil {
			log.Println(err)
		}
		fmt.Printf("consumer msg %v\n", result)

	}
}

List队列的缺点:

  1. 不支持重复消费:消费者拉取消息后,这条消息就从 List 中删除了,无法被其它消费者再次消费,即不支持多个消费者消费同一批数据,仅支持1对1。
  2. 消息丢失:消费者拉取到消息后,如果发生异常宕机,那这条消息就丢失了,无论消费者是否处理成功,这条消息都没办法再次消费了。

#发布/订阅模型 Pub/Sub

从名字就能看出来,这个模块是 Redis 专门是针对「发布/订阅」这种队列模型设计的。 它正好可以解决前面提到的第一个问题:重复消费。 即多组生产者、消费者的场景,我们来看它是如何做的。 Redis 提供了 PUBLISH / SUBSCRIBE 命令,来完成发布、订阅的操作。 image.png image.png 使用Pub/Sub就能很好的解决不能重复消费的问题,多个消费者能够消费同一个生产的消息。 除此之外,Pub/Sub 还提供了「匹配订阅」模式,允许消费者根据一定规则,订阅「多个」自己感兴趣的队列。 代码:

// 订阅

package main

import (
	"context"
	"fmt"
	"github.com/go-redis/redis/v8"
)


func main() {
	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "",
		DB:       1,
	})
	sub := rdb.Subscribe(context.TODO(), "queue")
	ch := sub.Channel()

	for msg := range ch {
		fmt.Println(msg.Channel, msg.Payload)
	}

}

// 发布
package main

import (
	"context"
	"github.com/go-redis/redis/v8"
)

func main() {
	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "",
		DB:       1,
	})
	data :=[]string{`{"name": "jinzhu11221212321", "age": 18, "tags": ["tag211", "tag2"], "orgs": {"orga": "orga"}`,`{"name": "asd", "age": 18, "tags": ["tag211", "tag2"], "orgs": {"orga": "orga"}`}
	for i:=range data {
		err := rdb.Publish(context.TODO(), "queue", data[i]).Err()
		if err != nil {
			panic(err)
		}
	}

}

Pub/Sub 最大问题是:丢数据。 如果发生以下场景,就有可能导致数据丢失:

  1. 消费者下线
  2. Redis 宕机
  3. 消息堆积

Pub/Sub的优缺点:

  1. 支持发布 / 订阅,支持多组生产者、消费者处理消息
  2. 消费者下线,数据会丢失
  3. 不支持数据持久化,Redis 宕机,数据也会丢失
  4. 消息堆积,缓冲区溢出,消费者会被强制踢下线,数据也会丢失

#Stream

首先,Stream 通过 XADD 和 XREAD 完成最简单的生产、消费模型:

生产者发布 2 条消息:

// *表示让Redis自动生成消息ID
127.0.0.1:6379> XADD queue * name zhangsan
"1618469123380-0"
127.0.0.1:6379> XADD queue * name lisi
"1618469127777-0"

使用 XADD 命令发布消息,其中的「*」表示让 Redis 自动生成唯一的消息 ID。 这个消息 ID 的格式是「时间戳-自增序号」。 消费者拉取消息:

// 从开头读取5条消息,0-0表示从开头读取
127.0.0.1:6379> XREAD COUNT 5 STREAMS queue 0-0
1) 1) "queue"
   2) 1) 1) "1618469123380-0"
         2) 1) "name"
            2) "zhangsan"
      2) 1) "1618469127777-0"
         2) 1) "name"
            2) "lisi"

如果想继续拉取消息,需要传入上一条消息的 ID:

127.0.0.1:6379> XREAD COUNT 5 STREAMS queue 1618469127777-0
(nil)

没有消息,Redis 会返回 NULL。

以上就是 Stream 最简单的生产、消费。

  1. Stream 是否支持「阻塞式」拉取消息? 可以的,在读取消息时,只需要增加 BLOCK 参数即可。
// BLOCK 0 表示阻塞等待,不设置超时时间
127.0.0.1:6379> XREAD COUNT 5 BLOCK 0 STREAMS queue 1618469127777-0

这时,消费者就会阻塞等待,直到生产者发布新的消息才会返回。 2) Stream 是否支持发布 / 订阅模式? 也没问题,Stream 通过以下命令完成发布订阅:

下面我们来看具体如何做? 首先,生产者依旧发布 2 条消息:

127.0.0.1:6379> XADD queue * name zhangsan
"1618470740565-0"
127.0.0.1:6379> XADD queue * name lisi
"1618470743793-0"

之后,我们想要开启 2 组消费者处理同一批数据,就需要创建 2 个消费者组:

// 创建消费者组1,0-0表示从头拉取消息
127.0.0.1:6379> XGROUP CREATE queue group1 0-0
OK
// 创建消费者组2,0-0表示从头拉取消息
127.0.0.1:6379> XGROUP CREATE queue group2 0-0
OK

消费者组创建好之后,我们可以给每个「消费者组」下面挂一个「消费者」,让它们分别处理同一批数据。 第一个消费组开始消费:

// group1的consumer开始消费,>表示拉取最新数据
127.0.0.1:6379> XREADGROUP GROUP group1 consumer COUNT 5 STREAMS queue >
1) 1) "queue"
   2) 1) 1) "1618470740565-0"
         2) 1) "name"
            2) "zhangsan"
      2) 1) "1618470743793-0"
         2) 1) "name"
            2) "lisi"

同样地,第二个消费组开始消费:

// group2的consumer开始消费,>表示拉取最新数据
127.0.0.1:6379> XREADGROUP GROUP group2 consumer COUNT 5 STREAMS queue >
1) 1) "queue"
   2) 1) 1) "1618470740565-0"
         2) 1) "name"
            2) "zhangsan"
      2) 1) "1618470743793-0"
         2) 1) "name"
            2) "lisi"

我们可以看到,这 2 组消费者,都可以获取同一批数据进行处理了。 这样一来,就达到了多组消费者「订阅」消费的目的。

  1. 消息处理时异常,Stream 能否保证消息不丢失,重新消费? 除了上面拉取消息时用到了消息 ID,这里为了保证重新消费,也要用到这个消息 ID。 当一组消费者处理完消息后,需要执行 XACK 命令告知 Redis,这时 Redis 就会把这条消息标记为「处理完成」。
// group1下的 1618472043089-0 消息已处理完成
127.0.0.1:6379> XACK queue group1 1618472043089-0

如果消费者异常宕机,肯定不会发送 XACK,那么 Redis 就会依旧保留这条消息。 待这组消费者重新上线后,Redis 就会把之前没有处理成功的数据,重新发给这个消费者。这样一来,即使消费者异常,也不会丢失数据了。

// 消费者重新上线,0-0表示重新拉取未ACK的消息
127.0.0.1:6379> XREADGROUP GROUP group1 consumer1 COUNT 5 STREAMS queue 0-0
// 之前没消费成功的数据,依旧可以重新消费
1) 1) "queue"
   2) 1) 1) "1618472043089-0"
         2) 1) "name"
            2) "zhangsan"
      2) 1) "1618472045158-0"
         2) 1) "name"
            2) "lisi"
  1. Stream 数据会写入到 RDB 和 AOF 做持久化吗? Stream 是新增加的数据类型,它与其它数据类型一样,每个写操作,也都会写入到 RDB 和 AOF 中。 我们只需要配置好持久化策略,这样的话,就算 Redis 宕机重启,Stream 中的数据也可以从 RDB 或 AOF 中恢复回来。
  2. 消息堆积时,Stream 是怎么处理的? 其实,当消息队列发生消息堆积时,一般只有 2 个解决方案:
  1. 生产者限流:避免消费者处理不及时,导致持续积压
  2. 丢弃消息:中间件丢弃旧消息,只保留固定长度的新消息

而 Redis 在实现 Stream 时,采用了第 2 个方案。 在发布消息时,你可以指定队列的最大长度,防止队列积压导致内存爆炸。

// 队列长度最大10000
127.0.0.1:6379> XADD queue MAXLEN 10000 * name zhangsan
"1618473015018-0"

当队列长度超过上限后,旧消息会被删除,只保留固定长度的新消息。 这么来看,Stream 在消息积压时,如果指定了最大长度,还是有可能丢失消息的。 代码:

package main

import (
	"context"
	"github.com/go-redis/redis/v8"
)

func main() {
	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "",
		DB:       1,
	})
	data := []string{`{"name": "jinzhu11221212321", "age": 18, "tags": ["tag211", "tag2"], "orgs": {"orga": "orga"}`, `{"name": "asd", "age": 18, "tags": ["tag211", "tag2"], "orgs": {"orga": "orga"}`}
	rdb.XAdd(context.TODO(), &redis.XAddArgs{
		Stream: "queue1",
		Values: data,
	})

}


package main

import (
	"context"
	"fmt"
	"github.com/go-redis/redis/v8"
)

func main() {
	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "",
		DB:       1,
	})
	result, err := rdb.XRead(context.TODO(), &redis.XReadArgs{
		Streams: []string{"queue1","0"},
		Count:   0,
		Block:   0,
	}).Result()
	if err != nil {
		panic(err)
	}
	for r:= range result {
		for m:=range result[r].Messages {
			fmt.Println(result[r].Messages[m].Values)
		}

	}

}


#参考

[1] https://redis.uptrace.dev/
[2] https://redis.io/commands
[3] https://mp.weixin.qq.com/s/RthQvzLHZRGNo-z6X_7jQQ
[4] https://github.com/overstarry/queue



/sw-load.js?v=e5ae5a1ed170f4499ac6292e7164b68528c51f6d6518cd75a49e6a6b737831d5728da21fc14dcbc7a91328e53858c6ff7195cc3fc8b25f0feeaef2af151d6686 /fireball.gif?v=569e393374f2af74d6c575090904aaf51e641e5eb5ea89ae7c7de01f7293abc165b3a7e8685690a8b951c778603fec98ae6822ff2f7ea86a536776966cb65d5d /favicon.ico?v=beac62000b1965d4e036575d58ef681f6d4c35c6b7ccafb1e286f99bebd3ca5f30f51dd9dcfb4b832132891ff814b18f1e040c08fc7a49be064016fab53c26b3 /favicon-16x16.png?v=5a9fcce4aea1dfb145b39a296c90c3fd0cac49dd7a83999a2bdd2a0ee4e6950f3a1b1f1fe14522b2fcb4cad75734f2a4e84fd964b56217748b9778a2e1697ff7 /favicon-32x32.png?v=35be3e52467cebe716e17b163f587373cd6c52d1993e868caae46dbfc53ba6955ca6ea2c77119ab6d8e535f2cdde502bcf7fe60984f83d2cdb36ee6b92ee37cb /icon-192x192.png?v=3820c1b1e6d755d2b7c2a04a65f0f1feef793b297f7ee995947137ccd8f73ec304457f6ce1df987a9a0a13ed7dacd203225505b832ccd2318b530ae53a55cebc /icon-512x512.png?v=de62ae905479fd813300d286ed1d2fe6bb6f6292623a5d918691642f6dd09a68943c69ed2a95a1820076919e69ff4fda668bb79e610ebc1d3200fedd7f634443 /apple-touch-icon.png?v=5d32464a608cc4b6e656e7be5bba48360b472b399ab82ebf4bbf4a93bb964e26f7bb0a1897ebbbdaf11444e7a93215f04f1c7fe8c08df5dddeeddbf97f93e149 /main.css?v=8ab3ba2ea49a89d1cee56b62a947c88597785d0842586199965b60438f9430279525aef9d01320077a32656df3e7435d48ccd7228cfc9bcd6a97ff0a4cc79358 /nerd-fonts.css?v=4213ecfcacca379b433c0fd135281c627c074e42d243cca41777df5738649704db63d448a19b13f80dfc9337485d8a1eb1a4b77cf2fa9d1fe2d3b6768c66e7bb /unstyle.css?v=b14bd48a2efbd463d973763aa3184c69aa02164c0891acacc9eab49ddd275f98f0050b4c31d2093e4671e7abe04f9459a041f0064384a90d97b8ff21b6824825 /langs.css?v=12474958ee314a9fde4704e1f5a032dc632d41f9461faca326ac284297766c4ceb07b45fec7fbc09fa72b0f21dcc64f0c31e64fc2e5e838b1d30f5fe540afd78 /syntax-theme-light.css?v=ccdddc2d2d88953c6d7d0376777b8409028ef625a7321dfa41619547b4f5eddbe89aa95ff5e7e2620da0ea13fbabebe2fd544620bc7e81e3294776b3425df48a /syntax-theme-dark.css?v=dfede4879841e4a58e5fc71115aa5f5b82e206d85eb771ff4e5a40a1d82621570aad2458f637365ae4370d9a1cf5070edc9765f7c2d4506e12e2ba3c6081ffd5 /sw-style.css?v=352cab856807e725351d62a9cae9dc445a675ab7e0bb0d3b12440b08dd574526c62827a5f4af706f7ad74df996a7f71f2c2a306fc1b188e1560007f0d4eda4fc /posts/page/2/ /posts/page/3/ /posts/page/4/ /posts/page/5/ /posts/page/6/ /posts/page/7/ /posts/page/8/ /posts/page/9/ /posts/page/10/ /posts/page/11/ /posts/page/12/ /posts/page/13/ /posts/page/14/ /posts/page/15/ /posts/page/16/ /categories/ /tags/ /tags/413/ /tags/a-li-yun/ /tags/a-li-yun-oss/ /tags/acme-sh/ /tags/adsense/ /tags/aes/ /tags/ai/ /tags/aliyun/ /tags/an-quan/ /tags/apisix/ /tags/archive-zip/ /tags/atop/ /tags/authing/ /tags/bei-fen/ /tags/ben-di-hua/ /tags/bian-ma/ /tags/bing-fa-bian-cheng/ /tags/bot/ /tags/buf/ /tags/casbin/ /tags/cdn/ /tags/ce-lue-mo-shi/ /tags/cert-manager/ /tags/certificatemanager/ /tags/chrome/ /tags/ci/ /tags/clarity/ /tags/clean-cache/ /tags/cody/ /tags/colab/ /tags/conc/ /tags/concurrency/ /tags/configmaps/ /tags/consul/ /tags/containerd/ /tags/coverage/ /tags/coze/ /tags/cpu/ /tags/crash/ /tags/crawler/ /tags/crypto/ /tags/cte/ /tags/cuo-wu-chu-li/ /tags/cve-2021-22205/ /tags/data-visualization/ /tags/database/ /tags/datax/ /tags/date/ /tags/decode/ /tags/dms/ /tags/dns/ /tags/dns-authorization/ /tags/docker/ /tags/duo-lu-fu-yong/ /tags/duo-ping-tai-bo-ke-fa-bu-gong-ju/ /tags/easeprobe/ /tags/email/ /tags/embed/ /tags/ent/ /tags/errgroup/ /tags/error/ /tags/external/ /tags/fang-wen-kong-zhi/ /tags/fen-bu-shi-lian-lu-zhui-zong/ /tags/ffmpeg/ /tags/finalizers/ /tags/fly-io/ /tags/fsck/ /tags/fu-wu-fan-she-xie-yi/ /tags/fu-zai-jun-heng/ /tags/gcloud/ /tags/gin/ /tags/github/ /tags/github-pages/ /tags/gitlab/ /tags/go/ /tags/golang/ /tags/gonew/ /tags/gong-ju/ /tags/google/ /tags/google-analytics/ /tags/google-api/ /tags/google-cloud/ /tags/google-oauth2/ /tags/govulncheck/ /tags/grafana/ /tags/grpc/ /tags/gzip/ /tags/health-check/ /tags/helm/ /tags/hosts/ /tags/http/ /tags/https/ /tags/hugo/ /tags/humanize/ /tags/i18n/ /tags/image-compress/ /tags/imap/ /tags/init/ /tags/jian-kang-jian-cha/ /tags/jian-kang-tan-zhen/ /tags/jian-kong/ /tags/json/ /tags/k8s/ /tags/katana/ /tags/ke-shi-hua/ /tags/kratos/ /tags/kubernetes/ /tags/lan-jie-qi/ /tags/lint/ /tags/linter/ /tags/linux/ /tags/liu-lan-qi/ /tags/load-balancing/ /tags/log/ /tags/loki/ /tags/lua/ /tags/magika/ /tags/mapping/ /tags/markdown/ /tags/memory/ /tags/mergo/ /tags/metabase/ /tags/microsoft/ /tags/minio/ /tags/mo-hu-ce-shi/ /tags/monitor/ /tags/nei-cun/ /tags/nginx/ /tags/nginx-ingress/ /tags/node-js/ /tags/novelai/ /tags/oauth2/ /tags/once/ /tags/opentelemetry/ /tags/opentracing/ /tags/openwrite/ /tags/os/ /tags/paas/ /tags/performance/ /tags/playwright/ /tags/playwright-go/ /tags/plugin/ /tags/png/ /tags/pngcrush/ /tags/pngquant/ /tags/postgresql/ /tags/profiling/ /tags/prometheus/ /tags/protobuf/ /tags/proxy/ /tags/psutil/ /tags/pyroscope/ /tags/rancher/ /tags/rand/ /tags/redis/ /tags/ren-gong-zhi-neng/ /tags/ren-zheng/ /tags/retry/ /tags/reverse-proxy/ /tags/rong-qi/ /tags/rueidis/ /tags/sealos/ /tags/security/ /tags/server-reflection/ /tags/serverless/ /tags/service/ /tags/she-ji-mo-shi/ /tags/shi-jian-chu-li/ /tags/shu-ju-fen-xi/ /tags/singleflight/ /tags/slug/ /tags/soap/ /tags/spider/ /tags/sql/ /tags/sqlc/ /tags/stable-diffusion/ /tags/storage/ /tags/superset/ /tags/swap/ /tags/sync/ /tags/tcp-udp/ /tags/template/ /tags/test/ /tags/text/ /tags/tianji/ /tags/time/ /tags/tls/ /tags/tong-xin-mo-shi/ /tags/trace/ /tags/trace-viewer/ /tags/traefik/ /tags/tu-pian/ /tags/ubuntu/ /tags/uri/ /tags/v0-dev/ /tags/video/ /tags/visualization/ /tags/visualstudio/ /tags/wang-luo/ /tags/wang-ye-tan-ce/ /tags/wasi/ /tags/wasm/ /tags/wire/ /tags/wireshark/ /tags/wsdl/ /tags/wsl/ /tags/xiao-wen-ti/ /tags/xie-cheng/ /tags/xun-huan/ /tags/you-jian/ /tags/yu-ming/ /tags/yuan-shu-ju/ /tags/zheng-shu/ /tags/zhong-jian-jian/ /tags/zhua-bao/ /tags/zhuang-tai-ma/ /posts/go-recivie-email/ /posts/go-soap-desc/ /posts/docker-desktop-proxy/ /posts/hugo-deploy-github/ /posts/v0dev/ /posts/markdown-preview-enhanced/ /posts/go-humanize-introduce/ /posts/wsl-error1/ /posts/cody/ /posts/katana/ /posts/tianji/ /posts/magika/ /posts/cozebot/ /posts/go-generate-slug/ /posts/metabase/ /posts/gogenerategaid/ /posts/docker-init-command/ /posts/grpc-client-load-balancing/ /posts/grpcqing-qiu-zhong-shi/ /posts/alliyunrdserr/ /posts/kubernetes-resource-reservation/ /posts/atop/ /posts/kubernetes-externalname/ /posts/go-refresh-cdn/ /posts/apisix-enabled-gzip/ /posts/ru-he-shou-ji-xi-tong-dang-ji-hou-de-nei-cun-zhuan-chu-xin-xi/ /posts/conversion-of-chinese-characters-into-pinyin/ /posts/mergo-desc/ /posts/conc-better-structured-concurrency-for-go/ /posts/kubernetes-externaltrafficpolicy/ /posts/clarity-learn/ /posts/postgresql-cte-expressions/ /posts/go-design-patterns-strategy/ /posts/go1-22-new-for-loop/ /posts/apisix-proxy-grpc-service/ /posts/golou-dong-guan-li-gong-ju-govulncheck/ /posts/go-wdsl/ /posts/nginx-ingress-httpqing-qiu-413wen-ti-ji-jie-jue-fang-fa/ /posts/playwright-gojian-jie/ /posts/google-api-go-clientdiao-yong-googleadsensebao-gao-jie-kou-shi-bai-de-wen-ti-ji-jie-jue-fang-an/ /posts/gonewjian-jie/ /posts/ent-sql-modifier/ /posts/golang-embedjian-dan-jie-shao/ /posts/nginxfan-xiang-dai-li-cuo-wu-de-wen-ti-ji-jie-jue-fang-fa/ /posts/shi-yong-cert-managershen-qing-mian-fei-zheng-shu/ /posts/go-i18n/ /posts/rueidisjian-jie/ /posts/postgresqlzen-me-jie-jue-division-by-zerowen-ti/ /posts/kubernetes-healthcheck/ /posts/ranchercattleclusteragentcouldnotresolvehost/ /posts/sealos-version-compare-error/ /posts/kubernetes-podxiu-gai-hostswen-jian/ /posts/apisixhu-lue-urida-xiao-xie/ /posts/sqlcchu-ti-yan/ /posts/apisix-dockerbu-shu-zhong-ding-xiang-wen-ti/ /posts/go-wasi/ /posts/goshi-xian-jian-dan-fan-xiang-dai-li/ /posts/certificatemanagershi-yong-dnsshou-quan-shen-qing-zheng-shu/ /posts/googlecloudqing-chu-cdnhuan-cun/ /posts/apisixshu-ju-bei-fen/ /posts/apisixru-he-tian-jia-zi-ding-yi-cha-jian/ /posts/apisixgen-ju-qing-qiu-hostfang-wen-bu-tong-lu-jing/ /posts/shi-yong-acmezi-dong-geng-xin-apisix-sslzheng-shu/ /posts/containerdben-di-diao-shi-huan-jing-da-jian/ /posts/containerdjian-dan-an-zhuang-he-ke-hu-duan-shi-yong/ /posts/helmjie-shao-ji-shi-yong/ /posts/ubuntu20-04she-zhi-dns/ /posts/postgresqlde-jsonlei-xing/ /posts/shi-yong-apisixdai-li-postgresqlfu-wu/ /posts/dataxshu-ju-tong-bu-zhong-yu-dao-de-wen-ti/ /posts/goshi-jian-chu-li-ku-carbon/ /posts/supersetjian-dan-shi-yong/ /posts/shi-yong-fly-iobu-shu-miniodui-xiang-cun-chu-fu-wu/ /posts/fly-iobu-shu-goying-yong/ /posts/pngya-suo-gong-ju/ /posts/apisixshi-xian-nginxde-proxy-hide-headercan-shu/ /posts/qian-hou-duan-shi-yong-aesjia-mi-chuan-shu-shu-ju/ /posts/fly-iochu-ti-yan/ /posts/aihui-hua-chu-ti-yan/ /posts/kubernetes-configmaps-subpath-no-reload/ /posts/easeprobejian-dan-jie-shao-shi-yong/ /posts/grpczhong-jian-jian/ /posts/gitlab-cve-2021-22205/ /posts/grpcqing-qiu-zhua-bao/ /posts/grpc-server-reflection/ /posts/prometheus-operato/ /posts/gojin-xing-liu-lan-qi-wang-ye-jie-tu/ /posts/gopsutiljie-shao/ /posts/ji-yi-ci-a-li-yun-ossbao-cuo-de-jie-jue/ /posts/grpcjian-kang-tan-zhen/ /posts/grpcjian-kang-jian-cha/ /posts/gofa-song-you-jian/ /posts/goxie-cheng-bi-bao-de-wen-ti/ /posts/goya-suo-pngtu-xiang-da-xiao/ /posts/gochu-li-zipjie-ya-luan-ma-wen-ti/ /posts/singleflightjie-shao/ /posts/goding-shi-jian-kong-httpszheng-shu/ /posts/gobing-fa-sync-oncejie-xi/ /posts/gojie-qu-shi-pin-mou-yi-zheng-tu-pian/ /posts/go-errgroup/ /posts/ying-yong-nei-cun-sheng-gao-yuan-yin-pai-cha/ /posts/duo-ping-tai-bo-ke-fa-bu-gong-ju-openwriteshi-yong/ /posts/bufchu-ru-men-2/ /posts/bufchu-ru-men-1/ /posts/gotong-ji-dai-ma-ce-shi-fu-gai-lu/ /posts/kswapd0-consumes-a-lot-of-cpu/ /posts/kswapd0xiao-hao-da-liang-cpu/ /posts/postgresqlxiu-gai-xu-lie-chan-sheng-qi-de-can-shu/ /posts/kratosye-wu-zhuang-tai-ma-he-httpzhuang-tai-ma-fen-chi/ /posts/apisixshi-yong-authingjin-xing-ren-zheng-deng-lu/ /posts/google-oauth2shi-jian/ /posts/gomo-hu-ce-shi/ /posts/pyroscope-chi-xu-fen-ping-tai/ /posts/fsck/ /posts/grpcdan-xiang-an-quan-lian-jie/ /posts/log-and-trace/ /posts/k8s-finalizers/ /posts/casbinxue-xi-1/ /posts/goru-he-shi-yong-si-you-cang-ku-mo-kuai/ /posts/grpcduo-lu-fu-yong/ /posts/trace-in-sql/ /posts/grpcyuan-shu-ju/ /posts/golangduo-ban-ben-gong-cun/ /posts/grpccuo-wu-chu-li/ /posts/grpclan-jie-qi/ /posts/grpctong-xin-mo-shi/ /posts/gochang-jian-linter/ /posts/wasmingo/ /posts/containerdpei-zhi-si-you-cang-ku/ /posts/rong-qi-chu-xian-shi-jian-yi-chang-wen-ti-ji-jie-jue-fang-fa/ /posts/consulxue-xi/ /posts/kubernetesan-zhuang-apisix/ /posts/she-zhi-rancherfu-wu-qi-de-ben-di-kubernetesji-qun/ /posts/shi-yong-sealosbu-shu-kubernetesji-qun/ /posts/kratoszi-ding-yi-handlerfunc-mei-you-qing-qiu-ri-zhi-de-wen-ti-ji-jie-jue/ /posts/rsstest/ /posts/golangsui-ji-timesleepchu-xian-de-wen-ti/ /posts/ru-he-zai-ginzhong-cha-kan-prometheuszhi-biao/ /posts/shi-yong-prometheusshou-ji-miniozhi-biao/ /posts/fen-bu-shi-lian-lu-zhui-zong-chu-tan-2/ /posts/fen-bu-shi-lian-lu-zhui-zong-chu-tan/ /posts/bian-li-maplie-biao-de-golangmo-ban/ /posts/golang-templates/ /posts/wireru-men/ /posts/gofan-xing-chu-tan/ /posts/functional-options/ /posts/docker-grafanaqi-dong-shi-bai/ /posts/nginxda-jian-jing-tai-tu-pian-zi-yuan-fu-wu-qi/ /posts/traefikru-men-shi-yong/ /posts/go-modulezhi-nan-he-chang-jian-wen-ti/ /posts/gitlab-cigou-jian-dockerjing-xiang/ /posts/github-pagezi-ding-yi-yu-ming/ /posts/gocuo-wu-shi-jian/ /posts/shi-yong-redisshi-xian-dui-lie/ /posts/my-first-post/ /atom.xml /posts/ c1tyh4ll.png?v=e6bb8cdead47e48c0deba1e0a3016070984b5f7271166a72638f9ec5a6ef2d2eb8012e8e4cb64f4f3b6574c6d708bf2ae660d04b8b59a6de675ce4d4d62dd4c3 bk-prk.png?v=b00246fb5faeab35a588f347224b00d53083a9d5f4ae8cd87c0c2e0432bf7f348c6a6ab6f4e4edaf5ddbf13ee34b24946dd0af4cd7db6f4f334599f38917ac9e bk-prk.png?v=b00246fb5faeab35a588f347224b00d53083a9d5f4ae8cd87c0c2e0432bf7f348c6a6ab6f4e4edaf5ddbf13ee34b24946dd0af4cd7db6f4f334599f38917ac9e /icon-192x192.png?v=3820c1b1e6d755d2b7c2a04a65f0f1feef793b297f7ee995947137ccd8f73ec304457f6ce1df987a9a0a13ed7dacd203225505b832ccd2318b530ae53a55cebc /sitemap.xml /search_index.en.json /search.js /elasticlunr.min.js?v=d106ab529e29f6be48a948124723fcf411e06b8e4ea4477b551f256d190991fe3ca7f121714ef8d9f594a4aa680f2bbd37a5d8004abfbf3ea6eb3d4ea259ec0f">