This project is the special configuration for AWS Elastic Beanstalk service. It uses empty java-8 service and installs proper ElasticSearch version during deployment stage.
Make sure you have installed
- AWS Beanstalk CLI Tools
$ eb initPlease remember that our default application name is elasticsearch-cluster
Few environment variables have to be specified before deployment
CLUSTER_NAME: This is a name of your new shiny ElasticSearch cluster, please avoid use ofelasticsearchnameAWS_ACCESS_KEY: AWS Key IDAWS_SECRET_KEY: AWS Secret KeyAWS_REGION: Region in which ES cluster will be createdEC2_TAG_NAME: This value should be equal to the AWS Name tag (same as environment name)MASTER_NODES: Amount of master nodes. The rule is simple, this number should equal to total number of nodes (N) divided by 2 plus 1.N / 2 + 1.PORT: Should always be set to 9200, unless you changed ES http port
See example in .env.example file
In order to create new cluster you need to execute following bash commands
$ ENV_VARS=$(cat .env | xargs | sed 's/ /,/g')
$ eb create -c 2pventures-elasticsearch-staging --envvars ${ENV_VARS} --platform=java-8 -i m3.large --scale 4 elasticsearch-stagingWhere 2pventures-elasticsearch-staging is a CNAME; elasticsearch-staging is the environment name; m3.large is a instance type and --scale 4 is how many nodes to create
After environment is created you need to change AWS Security Group rules. You have to allow all 9300-9400 TCP and ICMP traffic within SG group.
$ eb deploy elasticsearch-stagingWhere elasticsearch-staging is a environment name
- Do not scale down the cluster during peak times, this will cause cluster move shards around also some pending requests may fail
- When adding new nodes to cluster some requests may fail
- When added new nodes to cluster remember to change
MASTER_NODESvar