单线程序,基于IO多路复用,基于内存和c语言编写,性能高。redis官方命令
1 数据结构
1.1 key的层级
redis的key可以通过冒号(:)来划分层级,如下图mms:company:order,但系统中可以看到有不少没有的,这些都是不符合开发规则的,研发工程师偷懒。这些事情应该严格要求,否则留下来的人就会很麻烦,根本无法猜到这个key是从哪里来的。
1.2 数据结构
动画解析Redis为什么用跳表而不是红黑树?我看有些面试官就是很无趣,就好像高考阅读的出题人,总以为比写文章的人还理解,人家作者就是因为跳表比红黑树容易实现,性能也不差就用了,搞一堆题目为难面试者,搞得他们也能写出一个redis一样。
2 场景
3 k8s中部署redis集群
3.1 存储器
因为redis数据需要持久化,所以我认为删掉pvc的时候不应该把数据直接删掉,需要手工清理一下,故而单独给他配置了存储器
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:name: managed-redis-storage
provisioner: fuseim.pri/ifs
reclaimPolicy: Retain
mountOptions:- hard- nfsvers=4
3.2 部署集群
apiVersion: v1
kind: Namespace
metadata:name: middleware---
apiVersion: v1
kind: ConfigMap
metadata:name: redis-configmapnamespace: middlewarelabels:app: redis
data:redis.conf: |dir "/data"maxmemory 0maxmemory-policy volatile-lrumin-slaves-max-lag 5min-slaves-to-write 1rdbchecksum yesrdbcompression yesrepl-diskless-sync yessave 900 1save 300 10save 60 10000maxclients 10000sentinel.conf: |dir "/data"sentinel down-after-milliseconds mymaster 10000sentinel failover-timeout mymaster 180000sentinel parallel-syncs mymaster 5init.sh: |HOSTNAME="$(hostname)"INDEX="${HOSTNAME##*-}"MASTER="$(redis-cli -h redis -p 26379 sentinel get-master-addr-by-name mymaster | grep -E '[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}')"MASTER_GROUP="mymaster"QUORUM="2"REDIS_CONF=/data/conf/redis.confREDIS_PORT=6379SENTINEL_CONF=/data/conf/sentinel.confSENTINEL_PORT=26379SERVICE=redis-headlessset -eusentinel_update() {echo "Updating sentinel config"eval MY_SENTINEL_ID="\${SENTINEL_ID_$INDEX}"sed -i "1s/^/sentinel myid $MY_SENTINEL_ID\\n/" "$SENTINEL_CONF"sed -i "2s/^/sentinel monitor $MASTER_GROUP $1 $REDIS_PORT $QUORUM \\n/" "$SENTINEL_CONF"echo "sentinel announce-ip $ANNOUNCE_IP" >> $SENTINEL_CONFecho "sentinel announce-port $SENTINEL_PORT" >> $SENTINEL_CONF}redis_update() {echo "Updating redis config"echo "slaveof $1 $REDIS_PORT" >> "$REDIS_CONF"echo "slave-announce-ip $ANNOUNCE_IP" >> $REDIS_CONFecho "slave-announce-port $REDIS_PORT" >> $REDIS_CONF}copy_config() {cp /readonly-config/redis.conf "$REDIS_CONF"cp /readonly-config/sentinel.conf "$SENTINEL_CONF"}setup_defaults() {echo "Setting up defaults"if [ "$INDEX" = "0" ]; thenecho "Setting this pod as the default master"redis_update "$ANNOUNCE_IP"sentinel_update "$ANNOUNCE_IP"sed -i "s/^.*slaveof.*//" "$REDIS_CONF"elseDEFAULT_MASTER="$(getent hosts "redis-0.$SERVICE" | awk '{ print $1 }')"if [ -z "$DEFAULT_MASTER" ]; thenecho "Unable to resolve host"exit 1fiecho "Setting default slave config.."redis_update "$DEFAULT_MASTER"sentinel_update "$DEFAULT_MASTER"fi}find_master() {echo "Attempting to find master"if [ "$(redis-cli -h "$MASTER" ping)" != "PONG" ]; thenecho "Can't ping master, attempting to force failover"if redis-cli -h "$SERVICE" -p "$SENTINEL_PORT" sentinel failover "$MASTER_GROUP" | grep -q 'NOGOODSLAVE' ; then setup_defaultsreturn 0fisleep 10MASTER="$(redis-cli -h $SERVICE -p $SENTINEL_PORT sentinel get-master-addr-by-name $MASTER_GROUP | grep -E '[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}')"if [ "$MASTER" ]; thensentinel_update "$MASTER"redis_update "$MASTER"elseecho "Could not failover, exiting..."exit 1fielseecho "Found reachable master, updating config"sentinel_update "$MASTER"redis_update "$MASTER"fi}mkdir -p /data/conf/echo "Initializing config.."copy_configANNOUNCE_IP=$(getent hosts "redis-$INDEX.$SERVICE" | awk '{ print $1 }')if [ -z "$ANNOUNCE_IP" ]; then"Could not resolve the announce ip for this pod"exit 1elif [ "$MASTER" ]; thenfind_masterelsesetup_defaultsfiif [ "${AUTH:-}" ]; thenecho "Setting auth values"ESCAPED_AUTH=$(echo "$AUTH" | sed -e 's/[\/&]/\\&/g');sed -i "s/replace-default-auth/${ESCAPED_AUTH}/" "$REDIS_CONF" "$SENTINEL_CONF"fiecho "Ready..."---
apiVersion: v1
kind: ConfigMap
metadata:name: redis-probesnamespace: middlewarelabels:app: redis
data:check-quorum.sh: |#!/bin/shset -euMASTER_GROUP="mymaster"SENTINEL_PORT=26379REDIS_PORT=6379NUM_SLAVES=$(redis-cli -p "$SENTINEL_PORT" sentinel master mymaster | awk '/num-slaves/{getline; print}')MIN_SLAVES=1if [ "$1" = "$SENTINEL_PORT" ]; thenif redis-cli -p "$SENTINEL_PORT" sentinel ckquorum "$MASTER_GROUP" | grep -q NOQUORUM ; thenecho "ERROR: NOQUORUM. Sentinel quorum check failed, not enough sentinels found"exit 1fielif [ "$1" = "$REDIS_PORT" ]; thenif [ "$MIN_SLAVES" -gt "$NUM_SLAVES" ]; thenecho "Could not find enough replicating slaves. Needed $MIN_SLAVES but found $NUM_SLAVES"exit 1fifish /probes/readiness.sh "$1"readiness.sh: |#!/bin/shset -euCHECK_SERVER="$(redis-cli -p "$1" ping)"if [ "$CHECK_SERVER" != "PONG" ]; thenecho "Server check failed with: $CHECK_SERVER"exit 1fi---
apiVersion: v1
kind: ServiceAccount
metadata:name: redisnamespace: middlewarelabels:app: redis---
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:name: redisnamespace: middlewarelabels:app: redis
rules:
- apiGroups:- ""resources:- endpointsverbs:- get---
kind: RoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:name: redisnamespace: middlewarelabels:app: redis
subjects:
- kind: ServiceAccountname: redis
roleRef:apiGroup: rbac.authorization.k8s.iokind: Rolename: redis---
apiVersion: v1
kind: Service
metadata:name: redis-headlessnamespace: middlewarelabels:app: redis-haannotations:service.alpha.kubernetes.io/tolerate-unready-endpoints: "true"
spec:publishNotReadyAddresses: truetype: ClusterIPclusterIP: Noneports:- name: serverport: 6379protocol: TCPtargetPort: redis- name: sentinelport: 26379protocol: TCPtargetPort: sentinelselector:app: redis-ha---
apiVersion: v1
kind: Service
metadata:name: redisnamespace: middlewarelabels:app: redis-haannotations:
spec:type: ClusterIPports:- name: serverport: 6379protocol: TCPtargetPort: redis- name: sentinelport: 26379protocol: TCPtargetPort: sentinelselector:app: redis-ha---
apiVersion: apps/v1
kind: StatefulSet
metadata:name: redisnamespace: middlewarelabels:app: redis-ha
spec:selector:matchLabels:app: redis-haserviceName: redis-headlessreplicas: 3podManagementPolicy: OrderedReadyupdateStrategy:type: RollingUpdatetemplate:metadata:labels:app: redis-haspec:affinity:podAntiAffinity:requiredDuringSchedulingIgnoredDuringExecution:- labelSelector:matchLabels:app: redis-hatopologyKey: kubernetes.io/hostnamepreferredDuringSchedulingIgnoredDuringExecution:- weight: 100podAffinityTerm:labelSelector:matchLabels:app: redis-hatopologyKey: failure-domain.beta.kubernetes.io/zonesecurityContext:fsGroup: 1000runAsNonRoot: truerunAsUser: 1000serviceAccountName: redisimagePullSecrets:- name: 密码initContainers:- name: config-initimage: 10.101.10.2:8081/mid/redis:7.4.0#image: redis:latest # 此镜像也可用imagePullPolicy: IfNotPresentresources:{}command:- shargs:- /readonly-config/init.shenv:- name: SENTINEL_ID_0value: 0c09a3866dba0f3b43ef2e383b5dc05980900fd8- name: SENTINEL_ID_1value: e6be0f70406122877338f7c814b17a7c7b648d82- name: SENTINEL_ID_2value: 31f8f52b34feaddcabdd6bf1827aeb02be44d2e3volumeMounts:- name: configmountPath: /readonly-configreadOnly: true- name: datamountPath: /datacontainers:- name: redisimage: 10.101.10.2:8081/mid/redis:7.4.0imagePullPolicy: IfNotPresentcommand:- redis-serverargs:- /data/conf/redis.conflivenessProbe:exec:command: [ "sh", "/probes/readiness.sh", "6379"]initialDelaySeconds: 15periodSeconds: 5readinessProbe:exec:command: ["sh", "/probes/readiness.sh", "6379"]initialDelaySeconds: 15periodSeconds: 5resources:{}ports:- name: rediscontainerPort: 6379volumeMounts:- mountPath: /dataname: data- mountPath: /probesname: probes- name: sentinelimage: 10.101.10.2:8081/mid/redis:7.4.0imagePullPolicy: IfNotPresentcommand:- redis-sentinelargs:- /data/conf/sentinel.conflivenessProbe:exec:command: [ "sh", "/probes/readiness.sh", "26379"]initialDelaySeconds: 15periodSeconds: 5readinessProbe:exec:command: ["sh", "/probes/readiness.sh", "26379"]initialDelaySeconds: 15periodSeconds: 5resources:{}ports:- name: sentinelcontainerPort: 26379volumeMounts:- mountPath: /dataname: data- mountPath: /probesname: probesvolumes:- name: configconfigMap:name: redis-configmap- name: probesconfigMap:name: redis-probesvolumeClaimTemplates:- metadata:name: dataannotations:spec:accessModes:- "ReadWriteMany"resources:requests:storage: "5Gi"storageClassName: managed-redis-storage
3 Unable to resolve redis-headless.middleware.svc.cluster.local
查看系统日志发现错误日志如下:
2024-12-26 05:28:13.161 INFO com.alibaba.nacos.client.config.impl.ClientWorker - [fixed-dubbo-prod-nacos-headless.middleware.svc.cluster.local_8848] [publish-single] ok, dataId=com.whty.acc.voucher.api.service.VoucherDubboService:::provider:acc-voucher-server, group=dubbo, tenant=dubbo-prod, config={"annotations":[],"canonicalName":"com.whty.acc.voucher.api.service.VoucherDubboService","codeSource...
2024-12-26 10:10:50.045 ERROR org.redisson.connection.SentinelConnectionManager - Unable to resolve redis-headless.middleware.svc.cluster.local
io.netty.resolver.dns.DnsResolveContext$SearchDomainUnknownHostException: Failed to resolve 'redis-headless.middleware.svc.cluster.local' [A(1)] and search domain query for configured domains failed as well: [acc.svc.cluster.local, svc.cluster.local, cluster.local]at io.netty.resolver.dns.DnsResolveContext.finishResolve(DnsResolveContext.java:1088)at io.netty.resolver.dns.DnsResolveContext.tryToFinishResolve(DnsResolveContext.java:1035)at io.netty.resolver.dns.DnsResolveContext.query(DnsResolveContext.java:422)at io.netty.resolver.dns.DnsResolveContext.access$700(DnsResolveContext.java:66)
于是对比一下nacos和redis,为什么nacos是好使的,但redis却不行,如下图,出现了差异,那么这个差异是如何造成的呢?
[root@master8 ~]# kubectl exec -it acc-voucher-server-7dfc9b49-2t6bd -n acc -- nslookup nacos-headless.middleware.svc.cluster.local
nslookup: can't resolve '(null)': Name does not resolveName: nacos-headless.middleware.svc.cluster.local
Address 1: 10.42.0.29 10-42-0-29.nacos-web-service.middleware.svc.cluster.local
Address 2: 10.42.2.35 10-42-2-35.nacos-web-service.middleware.svc.cluster.local
Address 3: 10.42.1.21 10-42-1-21.nacos-web-service.middleware.svc.cluster.local
[root@master8 ~]# kubectl exec -it acc-voucher-server-7dfc9b49-2t6bd -n acc -- nslookup redis-headless.middleware.svc.cluster.local
nslookup: can't resolve '(null)': Name does not resolveName: redis-headless.middleware.svc.cluster.local
Address 1: 10.42.1.24 10-42-1-24.redis.middleware.svc.cluster.local
Address 2: 10.42.0.39 redis-0.redis-headless.middleware.svc.cluster.local
Address 3: 10.42.2.39 10-42-2-39.redis.middleware.svc.cluster.local