- CDK for deploying single-node and multi-node OpenSearch cluster with OpenSearch Dashboards
This project enables user to deploy either a single-node or a multi-node OpenSearch cluster. There are two stacks that get deployed:
- OpenSearch-Network-Stack: Use this stack to either use an existing Vpc or create a new Vpc. This stack also creates a new security group to manage access.
- OpenSearch-Infra-Stack: Sets up EC2 Auto-scaling group (ASG) (installs opensearch and opensearch-dashboards using userdata), cloudwatch logging, load balancer. Check your cluster log in the log group created from your stack in the cloudwatch.
- Requires NPM to be installed
- Install project dependencies using
npm installfrom this project directory - Configure aws credentials
In order to deploy both the stacks the user needs to provide a set of required and optional parameters listed below:
| Name | Requirement | Type | Description |
|---|---|---|---|
| distVersion | Required | string | The OpenSearch distribution version (released/un-released) the user wants to deploy |
| securityDisabled | Required | boolean | Enable or disable security plugin |
| adminPassword | Optionally required | string | This value is required when security plugin is enabled and the cluster version is greater or equal to 2.12.0
|
| minDistribution | Required | boolean | Is it the minimal OpenSearch distribution with no security and plugins |
| distributionUrl | Required | string | OpenSearch tar distribution url |
| cpuArch | Required | string | CPU platform for EC2, could be either x64 or arm64
|
| singleNodeCluster | Required | boolean | Set true for single-node cluster else false for multi-node |
| serverAccessType | Required | string | Restrict server access based on ip address (ipv4/ipv6), prefix list and/or security group. See Restricting Server Access for more details. |
| restrictServerAccessTo | Required | string | The value for serverAccessType, e.g., 10.10.10.10/32, pl-12345, sg-12345. See Restricting Server Access for more details. |
| dashboardsUrl | Optional | string | OpenSearch Dashboards tar distribution url |
| vpcId | Optional | string | Re-use existing vpc, provide vpc id |
| securityGroupId | Optional | boolean | Re-use existing security group, provide security group id |
| cidr | Optional | string | User provided CIDR block for new Vpc. Defaults to 10.0.0.0/16
|
| managerNodeCount | Optional | integer | Number of cluster manager nodes. Defaults to 3 |
| dataNodeCount | Optional | integer | Number of data nodes. Defaults to 2 |
| clientNodeCount | Optional | integer | Number of dedicated client nodes. Defaults to 0 |
| ingestNodeCount | Optional | integer | Number of dedicated ingest nodes. Defaults to 0 |
| mlNodeCount | Optional | integer | Number of dedicated machine learning nodes. Defaults to 0 |
| dataInstanceType | Optional | string | EC2 instance type for data node. Defaults to r5.xlarge. See options in lib/opensearch-config/node-config.ts for available options. E.g., -c dataInstanceType=m5.xlarge
|
| mlInstanceType | Optional | string | EC2 instance type for ml node. Defaults to r5.xlarge. See options in lib/opensearch-config/node-config.ts for available options. E.g., -c mlInstanceType=m5.xlarge
|
| jvmSysProps | Optional | string | A comma-separated list of key=value pairs that will be added to jvm.options as JVM system properties. |
| additionalConfig | Optional | string | Additional opensearch.yml config parameters passed as JSON. Please be aware that this JSON merges with original opensearch.yml overwriting duplicate keys e.g., --context additionalConfig='{"plugins.security.nodes_dn": ["CN=*.example.com, OU=SSL, O=Test, L=Test, C=DE", "CN=node.other.com, OU=SSL, O=Test, L=Test, C=DE"], "plugins.security.nodes_dn_dynamic_config_enabled": false}'
|
| additionalOsdConfig | Optional | string | Additional opensearch_dashboards.yml config parameters passed as JSON. Please be aware that this JSON merges with original opensearch-dashboards.yml overwriting duplicate keys. e.g., additionalOsdConfig='{"data.search.usageTelemetry.enabled": "true"}'
|
| suffix | Optional | string | An optional string identifier to be concatenated with infra stack name. |
| networkStackSuffix | Optional | string | An optional string identifier to be concatenated with network stack name. |
| region | Optional | string | User provided aws region |
| account | Optional | string | User provided aws account |
| dataNodeStorage | Optional | string | User provided ebs block storage size. Defaults to 100Gb |
| mlNodeStorage | Optional | string | User provided ebs block storage size. Defaults to 100Gb |
| use50PercentHeap | Optional | boolean | Boolean flag to use 50% of physical memory as heap. Defaults to 1GB. e.g., --context use50PercentHeap=true
|
| isInternal | Optional | boolean | Boolean flag to make network load balancer internal. Defaults to internet-facing e.g., --context isInternal=true
|
| enableRemoteStore | Optional | boolean | Boolean flag to enable Remote Store feature e.g., --context enableRemoteStore=true. See Enable Remote Store Feature for more details. Defaults to false |
| storageVolumeType | Optional | string | EBS volume type for all the nodes (data, ml, cluster manager). Defaults to gp2. See lib/opensearch-config/node-config.ts for available options. E.g., -c storageVolumeType=gp3. For SSD based instance (i.e. i3 family), it is used for root volume configuration. |
| customRoleArn | Optional | string | User provided IAM role arn to be used as ec2 instance profile. -c customRoleArn=arn:aws:iam::<AWS_ACCOUNT_ID>:role/<ROLE_NAME>
|
| customConfigFiles | Optional | string | You can provide an entire config file to be overwritten or added to OpenSearch and OpenSearch Dashboards. Pass string in the form of JSON with key as local path to the config file to read from and value as file on the server to overwrite/add. Note that the values in the JSON needs to have prefix of opensearch or opensearch-dashboards. Example: -c customConfigFiles='{"opensearch-config/config.yml": "opensearch/config/opensearch-security/config.yml", "opensearch-config/role_mapping.yml":"opensearch/config/opensearch-security/roles_mapping.yml", "/roles.yml": "opensearch/config/opensearch-security/roles.yml"}'
|
| enableMonitoring | Optional | boolean | Boolean flag to enable monitoring and alarms for Infra Stack. See InfraStackMonitoring class for more details. Defaults to false e.g., --context enableMonitoring=true
|
| certificateArn | Optional | string | Add ACM certificate to the any listener (OpenSearch or OpenSearch-Dashboards) whose port is mapped to 443. e.g., --context certificateArn=arn:1234
|
| mapOpensearchPortTo | Optional | integer | Load balancer port number to map to OpenSearch. e.g., --context mapOpensearchPortTo=8440 Defaults to 80 when security is disabled and 443 when security is enabled |
| mapOpensearchDashboardsPortTo | Optional | integer | Load balancer port number to map to OpenSearch-Dashboards. e.g., --context mapOpensearchDashboardsPortTo=443 Always defaults to 8443 |
-
Before starting this step, ensure that your AWS CLI is correctly configured with access credentials.
-
Also ensure that you're running these commands in the current directory
-
Next, install the required dependencies:
npm install
- First, run CDK bootstrap.
- You need to provide all the required context parameters in the command
cdk bootstrap aws://<aws-account-number>/<aws-region> --context securityDisabled=false \
--context minDistribution=false --context distributionUrl='https://artifacts.opensearch.org/releases/bundle/opensearch/2.3.0/opensearch-2.3.0-linux-x64.tar.gz' \
--context cpuArch='x64' --context singleNodeCluster=false --context dataNodeCount=3 \
--context dashboardsUrl='https://artifacts.opensearch.org/releases/bundle/opensearch-dashboards/2.3.0/opensearch-dashboards-2.3.0-linux-x64.tar.gz' \
--context distVersion=2.3.0 --context serverAccessType=ipv4 --context restrictServerAccessTo=10.10.10.10/32
- Now you are ready to synthesize the CloudFormation templates:
cdk synth "*" --context securityDisabled=false \
--context minDistribution=false --context distributionUrl='https://artifacts.opensearch.org/releases/bundle/opensearch/2.3.0/opensearch-2.3.0-linux-x64.tar.gz' \
--context cpuArch='x64' --context singleNodeCluster=false --context dataNodeCount=3 \
--context dashboardsUrl='https://artifacts.opensearch.org/releases/bundle/opensearch-dashboards/2.3.0/opensearch-dashboards-2.3.0-linux-x64.tar.gz' \
--context distVersion=2.3.0 --context serverAccessType=ipv4 --context restrictServerAccessTo=10.10.10.10/32
Please note that as of now we only support instances backed by Amazon Linux-2 amis.
cdk deploy "*" --context securityDisabled=false \
--context minDistribution=false --context distributionUrl='https://artifacts.opensearch.org/releases/bundle/opensearch/2.3.0/opensearch-2.3.0-linux-x64.tar.gz' \
--context cpuArch='x64' --context singleNodeCluster=false --context dataNodeCount=3 \
--context dashboardsUrl='https://artifacts.opensearch.org/releases/bundle/opensearch-dashboards/2.3.0/opensearch-dashboards-2.3.0-linux-x64.tar.gz' \
--context distVersion=2.3.0 --context serverAccessType=ipv4 --context restrictServerAccessTo=10.10.10.10/32
After CDK Stack deployment the user will be returned a load-balancer url which they can use to interact with the cluster.
curl -X GET "http://<load-balancer-url>/_cluster/health?pretty" for OpenSearch
To interact with dashboards use port 8443. Type http://<load-balancer-url>:8443 in your browser.
For security enabled cluster run curl -X GET https://<load-balancer-url> -u 'admin:admin' --insecure
The security enabled dashboard is accessible using http on port 8443
To restrict access please refer Client IP Preservation to restrict access on internet-facing network load balancer. You need to restrict access to your OpenSearch cluster endpoint (load balancer).
Below values are allowed:
| serverAccessType | restrictServerAccessTo |
|---|---|
| ipv4 | all (0.0.0.0/0) or any ipv4 CIDR (eg: 10.10.10.10/32) |
| ipv6 | all (::/0) or any ipv6 CIDR (eg: 2001:0db8:85a3:0000:0000:8a2e:0370:7334) |
| prefixList | Prefix List id (eg: ab-12345) |
| securityGroupId | A security group ID (eg: sg-123456789) |
Remote Store feature provides an option to store indexed data in a remote durable data store. To enable this feature the user needs to register a snapshot repository (S3 or File System) which is used to store the index data.
Apart from passing enableRemoteStore flag as true the user needs to be provide additional settings to opensearch.yml, the settings are:
1. opensearch.experimental.feature.remote_store.enabled: 'true'
2. cluster.remote_store.enabled: 'true'
3. opensearch.experimental.feature.segment_replication_experimental.enabled: 'true'
4. cluster.indices.replication.strategy: SEGMENT
The above-mentioned settings need to be passed using additionalConfig parameter.
Please note the experimental settings are only applicable till the feature is under development and will be removed when the feature becomes GA.
The opensearch logs are available in cloudwatch logs log-group opensearchLogGroup/opensearch.log in the same region your stack is deployed.
Each EC2 instance will create its own log-stream and the log-stream will be named after each instance-id.
All the ec2 instances are hosted in private subnet and can only be accessed using AWS Systems Manager Session Manager
The ports to access the cluster are dependent on the security parameter value
- If
securityisdisable(HTTP),- OpenSearch 9200 is mapped to port 80 on the LB
- If
securityisenable(HTTPS),- OpenSearch 9200 is mapped to port 443 on the LB
- OpenSearch-Dasboards 5601 is always mapped to port 8443 on the LB (HTTP)
Change the port mapping using context variables
mapOpensearchPortToandmapOpensearchDashboardsPortTo. See available context parameters) for more details
To delete a particular stack use the command:
cdk destroy <stackName> <pass all the context parameters>
To delete all the created stacks together use the command:
cdk destroy --all <pass all the context parameters>
See developer guide and how to contribute to this project.
If you find a bug, or have a feature request, please don't hesitate to open an issue in this repository.
For more information, see project website and documentation. If you need help and are unsure where to open an issue, try forums.
This project has adopted the Amazon Open Source Code of Conduct. For more information see the Code of Conduct FAQ, or contact opensource-codeofconduct@amazon.com with any additional questions or comments.
See CONTRIBUTING for more information.
This project is licensed under the Apache-2.0 License.