Skip to content

Latest commit

 

History

History
131 lines (96 loc) · 3.4 KB

README.md

File metadata and controls

131 lines (96 loc) · 3.4 KB
Raw

Ambari-Redis-Service

Project

Redis 5.0.x

Feature Description
Version 5.0.0
Service Redis
Component-Master Redis-Master
Component-Slave Redis-Slave

Prerequisite

  1. Internet Access (need to yum install ruby etc);
  2. Ambari 2.x (test on 2.6.2.0) fully installed.

Features

  1. Support redis single instance deployment: 1 instance;
  2. Support redis cluster mode deployment: 3+ master instance and 3+ slave instance.

Installation

  1. Download the source code.
  2. Copy the project folder to ambari 's lib folder. Such as:
cp -r ./REDIS-5.0.x /var/lib/ambari-server/resources/stacks/HDP/2.6/services
  1. Restart ambari-server:
ambari-server restart

  1. Log into web then add service as usual: add redis service

  2. Configure the service installation:

For single instance deployment

1). choose 1 node be the master: config master

2). choose 0 node be the slave: config slave

3). configure down load url: config download url

4). configure cluster disabled: config cluster disabled

For cluster mode deployment

1). choose at least 3 nodes be the master: config master

2). choose at least 3 nodes be the slave: config slave

3). configure down load url: config download url

4). configure cluster enabled: config cluster enabled

5). configure network: config network

  1. Then continue the deployment and enjoy urself.^^

Background Maintaince

Here is some manually maintaince method specially for cluster operations (All the {} pamas can be found in Redis Configs in Ambari Web):

Check cluster:

{redis.base.dir}/src/redis-cli --cluster check {host_ip}:{port}

Create cluster:

{redis.base.dir}/src/redis-cli --cluster create host1_ip:{master-port} host2_ip:{master-port} host3_ip:{master-port} ...

Add master node to cluster:

{redis.base.dir}/src/redis-cli --cluster add-node {new_host_ip}:{master-port} {exist_cluster_host_ip}:{master-port}
{redis.base.dir}/src/redis-cli --cluster reshard {new_host_ip}:{master-port}

Add slave node to cluster:

{redis.base.dir}/src/redis-cli --cluster add-node {new_host_ip}:{master-port} {exist_cluster_host_ip}:{slave-port} --cluster-slave

Remove node from cluster:

# get node_id by check or info 
{redis.base.dir}/src/redis-cli --cluster check {host_ip}:{master-port}

# then remove
{redis.base.dir}/src/redis-cli --cluster del-node {host_ip}:{port} {node_id}

Clean up node:

# if node is already removed from cluster and stopped.
rm -fr {redis.base.dir} && rm -fr {log.file.dir} && rm -fr {pid.file.dir} && rm -fr {dir}

Re add a removed node back to cluster:

# remove the nodes.conf in dir
rm -fr {dir}/nodes.conf

# start redis (can be done on web)
{redis.base.dir}/src/redis-server {redis.base.dir}/redis.conf

# cleanup db
{redis.base.dir}/src/redis-cli
> flushdb

# use method above-mentioned add node to master or slave

Notice

  1. Only support creating master node cluster for the first time.
  2. Master node scaling need to be done mannually.
  3. Support salve node incremental scaling.
  4. Need more test if using on production.