|
| 1 | +# Setup Service to provide access to Zookeeper for clients |
| 2 | +apiVersion: v1 |
| 3 | +kind: Service |
| 4 | +metadata: |
| 5 | + # DNS would be like zookeeper.zoons |
| 6 | + name: zookeeper |
| 7 | + labels: |
| 8 | + app: zookeeper |
| 9 | +spec: |
| 10 | + ports: |
| 11 | + - port: 2181 |
| 12 | + name: client |
| 13 | + - port: 7000 |
| 14 | + name: prometheus |
| 15 | + selector: |
| 16 | + app: zookeeper |
| 17 | + what: node |
| 18 | +--- |
| 19 | +# Setup Headless Service for StatefulSet |
| 20 | +apiVersion: v1 |
| 21 | +kind: Service |
| 22 | +metadata: |
| 23 | + # DNS would be like zookeeper-0.zookeepers.etc |
| 24 | + name: zookeepers |
| 25 | + labels: |
| 26 | + app: zookeeper |
| 27 | +spec: |
| 28 | + ports: |
| 29 | + - port: 2888 |
| 30 | + name: server |
| 31 | + - port: 3888 |
| 32 | + name: leader-election |
| 33 | + clusterIP: None |
| 34 | + selector: |
| 35 | + app: zookeeper |
| 36 | + what: node |
| 37 | +--- |
| 38 | +# Setup max number of unavailable pods in StatefulSet |
| 39 | +apiVersion: policy/v1 |
| 40 | +kind: PodDisruptionBudget |
| 41 | +metadata: |
| 42 | + name: zookeeper-pod-disruption-budget |
| 43 | +spec: |
| 44 | + selector: |
| 45 | + matchLabels: |
| 46 | + app: zookeeper |
| 47 | + maxUnavailable: 1 |
| 48 | +--- |
| 49 | +# Setup Zookeeper StatefulSet |
| 50 | +# Possible params: |
| 51 | +# 1. replicas |
| 52 | +# 2. memory |
| 53 | +# 3. cpu |
| 54 | +# 4. storage |
| 55 | +# 5. storageClassName |
| 56 | +# 6. user to run app |
| 57 | +apiVersion: apps/v1 |
| 58 | +kind: StatefulSet |
| 59 | +metadata: |
| 60 | + # nodes would be named as zookeeper-0, zookeeper-1, zookeeper-2 |
| 61 | + name: zookeeper |
| 62 | + labels: |
| 63 | + app: zookeeper |
| 64 | +spec: |
| 65 | + selector: |
| 66 | + matchLabels: |
| 67 | + app: zookeeper |
| 68 | + serviceName: zookeepers |
| 69 | + replicas: 1 |
| 70 | + updateStrategy: |
| 71 | + type: RollingUpdate |
| 72 | + podManagementPolicy: OrderedReady |
| 73 | + template: |
| 74 | + metadata: |
| 75 | + labels: |
| 76 | + app: zookeeper |
| 77 | + what: node |
| 78 | + annotations: |
| 79 | + prometheus.io/port: '7000' |
| 80 | + prometheus.io/scrape: 'true' |
| 81 | + spec: |
| 82 | + affinity: |
| 83 | + podAntiAffinity: |
| 84 | + requiredDuringSchedulingIgnoredDuringExecution: |
| 85 | + - labelSelector: |
| 86 | + matchExpressions: |
| 87 | + - key: "app" |
| 88 | + operator: In |
| 89 | + values: |
| 90 | + - zookeeper |
| 91 | + # TODO think about multi-AZ EKS |
| 92 | + # topologyKey: topology.kubernetes.io/zone |
| 93 | + topologyKey: "kubernetes.io/hostname" |
| 94 | + containers: |
| 95 | + - name: kubernetes-zookeeper |
| 96 | + imagePullPolicy: IfNotPresent |
| 97 | + image: "docker.io/zookeeper:3.8.3" |
| 98 | + resources: |
| 99 | + requests: |
| 100 | + memory: "512M" |
| 101 | + cpu: "1" |
| 102 | + limits: |
| 103 | + memory: "4Gi" |
| 104 | + cpu: "2" |
| 105 | + ports: |
| 106 | + - containerPort: 2181 |
| 107 | + name: client |
| 108 | + - containerPort: 2888 |
| 109 | + name: server |
| 110 | + - containerPort: 3888 |
| 111 | + name: leader-election |
| 112 | + - containerPort: 7000 |
| 113 | + name: prometheus |
| 114 | + env: |
| 115 | + - name: SERVERS |
| 116 | + value: "1" |
| 117 | + |
| 118 | +# See those links for proper startup settings: |
| 119 | +# https://github.com/kow3ns/kubernetes-zookeeper/blob/master/docker/scripts/start-zookeeper |
| 120 | +# https://clickhouse.yandex/docs/en/operations/tips/#zookeeper |
| 121 | +# https://github.com/ClickHouse/ClickHouse/issues/11781 |
| 122 | + command: |
| 123 | + - bash |
| 124 | + - -x |
| 125 | + - -c |
| 126 | + - | |
| 127 | + HOST=`hostname -s` && |
| 128 | + DOMAIN=`hostname -d` && |
| 129 | + CLIENT_PORT=2181 && |
| 130 | + SERVER_PORT=2888 && |
| 131 | + ELECTION_PORT=3888 && |
| 132 | + PROMETHEUS_PORT=7000 && |
| 133 | + ZOO_DATA_DIR=/var/lib/zookeeper/data && |
| 134 | + ZOO_DATA_LOG_DIR=/var/lib/zookeeper/datalog && |
| 135 | + { |
| 136 | + echo "clientPort=${CLIENT_PORT}" |
| 137 | + echo 'tickTime=2000' |
| 138 | + echo 'initLimit=300' |
| 139 | + echo 'syncLimit=10' |
| 140 | + echo 'maxClientCnxns=2000' |
| 141 | + echo 'maxTimeToWaitForEpoch=2000' |
| 142 | + echo 'maxSessionTimeout=60000000' |
| 143 | + echo "dataDir=${ZOO_DATA_DIR}" |
| 144 | + echo "dataLogDir=${ZOO_DATA_LOG_DIR}" |
| 145 | + echo 'autopurge.snapRetainCount=10' |
| 146 | + echo 'autopurge.purgeInterval=1' |
| 147 | + echo 'preAllocSize=131072' |
| 148 | + echo 'snapCount=3000000' |
| 149 | + echo 'leaderServes=yes' |
| 150 | + echo 'standaloneEnabled=false' |
| 151 | + echo '4lw.commands.whitelist=*' |
| 152 | + echo 'metricsProvider.className=org.apache.zookeeper.metrics.prometheus.PrometheusMetricsProvider' |
| 153 | + echo "metricsProvider.httpPort=${PROMETHEUS_PORT}" |
| 154 | + echo "skipACL=true" |
| 155 | + echo "fastleader.maxNotificationInterval=10000" |
| 156 | + } > /conf/zoo.cfg && |
| 157 | + { |
| 158 | + echo "zookeeper.root.logger=CONSOLE" |
| 159 | + echo "zookeeper.console.threshold=INFO" |
| 160 | + echo "log4j.rootLogger=\${zookeeper.root.logger}" |
| 161 | + echo "log4j.appender.CONSOLE=org.apache.log4j.ConsoleAppender" |
| 162 | + echo "log4j.appender.CONSOLE.Threshold=\${zookeeper.console.threshold}" |
| 163 | + echo "log4j.appender.CONSOLE.layout=org.apache.log4j.PatternLayout" |
| 164 | + echo "log4j.appender.CONSOLE.layout.ConversionPattern=%d{ISO8601} - %-5p [%t:%C{1}@%L] - %m%n" |
| 165 | + } > /conf/log4j.properties && |
| 166 | + echo 'JVMFLAGS="-Xms128M -Xmx4G -XX:ActiveProcessorCount=8 -XX:+AlwaysPreTouch -Djute.maxbuffer=8388608 -XX:MaxGCPauseMillis=50"' > /conf/java.env && |
| 167 | + if [[ $HOST =~ (.*)-([0-9]+)$ ]]; then |
| 168 | + NAME=${BASH_REMATCH[1]} && |
| 169 | + ORD=${BASH_REMATCH[2]}; |
| 170 | + else |
| 171 | + echo "Failed to parse name and ordinal of Pod" && |
| 172 | + exit 1; |
| 173 | + fi && |
| 174 | + mkdir -pv ${ZOO_DATA_DIR} && |
| 175 | + mkdir -pv ${ZOO_DATA_LOG_DIR} && |
| 176 | + whoami && |
| 177 | + chown -Rv zookeeper "$ZOO_DATA_DIR" "$ZOO_DATA_LOG_DIR" && |
| 178 | + export MY_ID=$((ORD+1)) && |
| 179 | + echo $MY_ID > $ZOO_DATA_DIR/myid && |
| 180 | + for (( i=1; i<=$SERVERS; i++ )); do |
| 181 | + echo "server.$i=$NAME-$((i-1)).$DOMAIN:$SERVER_PORT:$ELECTION_PORT" >> /conf/zoo.cfg; |
| 182 | + done && |
| 183 | + if [[ $SERVERS -eq 1 ]]; then |
| 184 | + echo "group.1=1" >> /conf/zoo.cfg; |
| 185 | + else |
| 186 | + echo "group.1=1:2:3" >> /conf/zoo.cfg; |
| 187 | + fi && |
| 188 | + for (( i=1; i<=$SERVERS; i++ )); do |
| 189 | + WEIGHT=1 |
| 190 | + if [[ $i == 1 ]]; then |
| 191 | + WEIGHT=10 |
| 192 | + fi |
| 193 | + echo "weight.$i=$WEIGHT" >> /conf/zoo.cfg; |
| 194 | + done && |
| 195 | + zkServer.sh start-foreground |
| 196 | + readinessProbe: |
| 197 | + exec: |
| 198 | + command: |
| 199 | + - bash |
| 200 | + - -c |
| 201 | + - ' |
| 202 | + IFS=; |
| 203 | + MNTR=$(exec 3<>/dev/tcp/127.0.0.1/2181 ; printf "mntr" >&3 ; tee <&3; exec 3<&- ;); |
| 204 | + while [[ "$MNTR" == "This ZooKeeper instance is not currently serving requests" ]]; |
| 205 | + do |
| 206 | + echo "wait mntr works"; |
| 207 | + sleep 1; |
| 208 | + MNTR=$(exec 3<>/dev/tcp/127.0.0.1/2181 ; printf "mntr" >&3 ; tee <&3; exec 3<&- ;); |
| 209 | + done; |
| 210 | + STATE=$(echo -e $MNTR | grep zk_server_state | cut -d " " -f 2); |
| 211 | + if [[ "$STATE" =~ "leader" ]]; then |
| 212 | + echo "check leader state"; |
| 213 | + SYNCED_FOLLOWERS=$(echo -e $MNTR | grep zk_synced_followers | awk -F"[[:space:]]+" "{print \$2}" | cut -d "." -f 1); |
| 214 | + if [[ "$SYNCED_FOLLOWERS" != "0" ]]; then |
| 215 | + ./bin/zkCli.sh ls /; |
| 216 | + exit $?; |
| 217 | + else |
| 218 | + exit 0; |
| 219 | + fi; |
| 220 | + elif [[ "$STATE" =~ "follower" ]]; then |
| 221 | + echo "check follower state"; |
| 222 | + PEER_STATE=$(echo -e $MNTR | grep zk_peer_state); |
| 223 | + if [[ "$PEER_STATE" =~ "following - broadcast" ]]; then |
| 224 | + ./bin/zkCli.sh ls /; |
| 225 | + exit $?; |
| 226 | + else |
| 227 | + exit 1; |
| 228 | + fi; |
| 229 | + else |
| 230 | + exit 1; |
| 231 | + fi |
| 232 | + ' |
| 233 | + initialDelaySeconds: 10 |
| 234 | + periodSeconds: 60 |
| 235 | + timeoutSeconds: 60 |
| 236 | + livenessProbe: |
| 237 | + exec: |
| 238 | + command: |
| 239 | + - bash |
| 240 | + - -xc |
| 241 | + - 'date && OK=$(exec 3<>/dev/tcp/127.0.0.1/2181 ; printf "ruok" >&3 ; IFS=; tee <&3; exec 3<&- ;); if [[ "$OK" == "imok" ]]; then exit 0; else exit 1; fi' |
| 242 | + initialDelaySeconds: 10 |
| 243 | + periodSeconds: 30 |
| 244 | + timeoutSeconds: 5 |
| 245 | + volumeMounts: |
| 246 | + - name: datadir-volume |
| 247 | + mountPath: /var/lib/zookeeper |
| 248 | + # Run as a non-privileged user |
| 249 | + securityContext: |
| 250 | + runAsUser: 1000 |
| 251 | + fsGroup: 1000 |
| 252 | + volumeClaimTemplates: |
| 253 | + - metadata: |
| 254 | + name: datadir-volume |
| 255 | + spec: |
| 256 | + accessModes: |
| 257 | + - ReadWriteOnce |
| 258 | + resources: |
| 259 | + requests: |
| 260 | + storage: 25Gi |
0 commit comments