jaychanggithub / k8s-redis-cluster

Redis Cluster setup running on Kubernetes.

Cluster-NoSQL-Redis

---

Suitable Project

• None

---

Version

Rev: 1.0.2

---

Status

---

Description

• Base on kubernetes cluster ecosystem to build up the redis HA cluster.
• The purpose to deployment redis-tool is used to sync data, backup data and restore data.
• Use cronjob object of kubernetes to backup/dump data every hour.

---

Usage

• Redis cluster instance.

  redis_pods = [
      'redis-app-0',
      'redis-app-1',
      'redis-app-2',
      'redis-app-3',
      'redis-app-4',
      'redis-app-5'
  ]

• Execute the container from image of redis-tool.

  $ kubectl run -it redis-tool \
          --rm=true \
          --image="registry.ipt-gitlab:8081/sit-develop-tool/cluster-nosql-redis/redis-tool:$VERSION" \
          --overrides='{ "apiVersion": "v1", "spec": { "imagePullSecrets": [{"name": "gitlab-registry"}] } }' \
          --restart=Never bash -n kube-ops

• Using python shell to list data keys.

  $ kubectl run -it redis-tool \
          --rm=true \
          --image="registry.ipt-gitlab:8081/sit-develop-tool/cluster-nosql-redis/redis-tool:$VERSION" \
          --overrides='{ "apiVersion": "v1", "spec": { "imagePullSecrets": [{"name": "gitlab-registry"}] } }' \
          --restart=Never bash -n kube-ops
  
  $ python
  
  
  import sys
  from rediscluster import RedisCluster
  def init_redis():
      startup_nodes = [
          {'host': 'redis-app-0.redis-headless.kube-ops.svc.cluster.local', 'port': 6379},
          {'host': 'redis-app-1.redis-headless.kube-ops.svc.cluster.local', 'port': 6379},
          {'host': 'redis-app-2.redis-headless.kube-ops.svc.cluster.local', 'port': 6379},
          {'host': 'redis-app-3.redis-headless.kube-ops.svc.cluster.local', 'port': 6379},
          {'host': 'redis-app-4.redis-headless.kube-ops.svc.cluster.local', 'port': 6379},
          {'host': 'redis-app-5.redis-headless.kube-ops.svc.cluster.local', 'port': 6379}
      ]
      try:
          conn = RedisCluster(startup_nodes=startup_nodes,
                              skip_full_coverage_check=True,
                              decode_responses=True)
          print('连接成功！！！！！1', conn)
          return conn
      except Exception as e:
          print("connect error ", str(e))
          sys.exit(1)
  
  rc = init_redis()
  rc.scan()

• To dump the redis server data.

  # The first startup the container
  $ kubectl run -it redis-tool \
          --rm=true \
          --image="registry.ipt-gitlab:8081/sit-develop-tool/cluster-nosql-redis/redis-tool:$VERSION" \
          --overrides='{ "apiVersion": "v1", "spec": { "imagePullSecrets": [{"name": "gitlab-registry"}] } }' \
          --restart=Never bash -n kube-ops
  
  # When the redis-tool container was running
  $ kubectl exec -it redis-tool -n kube-ops bash
  
  
  # dump the cluster instance
  $ python redis-tool.py -n kube-ops -st cluster -m dump -p 6379
  
  # dump the standalone instance
  $ python redis-tool.py -n kube-ops -st standalone -s <instance_redis_ip> -m dump -p 6379
  
  # check dump data
  $ ls -al ./reports/$timestemp/

• To restore the redis from specified data.

  # The first startup the container
  $ kubectl run -it redis-tool \
          --rm=true \
          --image="registry.ipt-gitlab:8081/sit-develop-tool/cluster-nosql-redis/redis-tool:$VERSION" \
          --overrides='{ "apiVersion": "v1", "spec": { "imagePullSecrets": [{"name": "gitlab-registry"}] } }' \
          --restart=Never bash -n kube-ops
  
  # When the redis-tool container was running
  $ kubectl exec -it redis-tool -n kube-ops bash
  
  # Restore the redis from specified data
  $ python redis-tool.py -n kube-ops -dt <standalone|cluster> -p 6379 -d <cluster_ignore|10.99.104.251> -m restore -rdir /usr/src/reports/$timestemp
  
  
  # check log showing as follow, press 'Ctrl + C'
  $ cat reports/redis-shake-restore.log
  2020/11/02 14:03:17 [INFO] restore from '[/usr/src/reports/20201102140026/dump-20201102140026.0 /usr/src/reports/20201102140026/dump-20201102140026.1 /usr/src/reports/20201102140026/dump-20201102140026.2]' to '[10.99.104.251:6379]' done
  2020/11/02 14:03:17 [INFO] Enabled http stats, set status (incr), and wait forever.

• To synchronize the end-to-end redis instance

  # The first startup the container
  $ kubectl run -it redis-tool \
          --rm=true \
          --image="registry.ipt-gitlab:8081/sit-develop-tool/cluster-nosql-redis/redis-tool:$VERSION" \
          --overrides='{ "apiVersion": "v1", "spec": { "imagePullSecrets": [{"name": "gitlab-registry"}] } }' \
          --restart=Never bash -n kube-ops
  
  # When the redis-tool container was running
  $ kubectl exec -it redis-tool -n kube-ops bash
  
  
  # redis cluster to redis standalone(10.99.104.251)
  $ python redis-tool.py -p 6379 -n kube-ops -m sync
  
  # redis standalone instance(10.99.104.251) to cluster
  $ python redis-tool.py -p 6379 -n kube-ops -tt cluster  -s 10.99.104.251 -st standalone -m sync

---

Log

• The dump log is reports/redis-shake-dump.log
• The sync log is reports/redis-shake-sync.log
• The restore log is reports/redis-shake-restore.log

---

Contact

Author: Chang.Jay