Skip to content

Creates a complete Couchbase cluster in AWS, with support for Multidimensional Scaling

License

Notifications You must be signed in to change notification settings

brantburnett/terraform-aws-couchbase

Repository files navigation

terraform-aws-couchbase

Terraform module for creating a Couchbase cluster.

How To Use

The scripts may be referenced using Terraform Modules. For details on modules, see the Terraform documentation. The module may be referenced directly from GitHub, or you can make a local copy into a subfolder.

For a variable reference, see the variables.tf file.

Example

There is an example available in the example folder.

Multidimensional Scaling Groups

Please see the Couchbase Documentation on Multidimensional Scaling for more information on the concept.

Up to five MDS groups can be defined. Each MDS group may run different services, have different node types, etc. Information about each MDS group is supplied via map variables, with the groups numbered 1 through 5.

Only MDS groups with a node_count greater than 0 are created. Each MDS group must also have a unique name.

node_count = {
    "1" = 2
    "2" = 1
    "3" = 1
}

group_name = {
    "1" = "Data"
    "2" = "Index"
    "3" = "Query"
}

# It is valid to include more than one service in a single MDS group
# However, the first MDS group should at least include the data service
# Note: If you use Community Edition, you must use a single MDS group with all services.

services = {
    "1" = ["data"]
    "2" = ["index"]
    "3" = ["query"]
}

For most other MDS group settings, such as instance_type, if you set the value for group "1" it will be automatically used for all other groups unless expressly overridden.

AMI

This module requires the use of an Amazon Linux AMI. By default, the latest Amazon-provided AMI is used automatically. However, you may supply the ami variable to override this behavior and use a custom AMI, so long as it is still based on Amazon Linux.

Data Volume

To control the size and type of your data volume, supply a data volume setting via data_volume. This setting is per MDS group, but the value for group 1 will be the default for all other groups.

data_volume = {
    "1" = {
        volume_type = "gp2" # SSD
        volume_size = "100" # Size is in GB
    }
}

You may supply any settings available for an ebs_block_device, except for device_name.

Note that this setting only takes effect for new clusters and newly launched nodes, not existing nodes.

Security Groups

Security groups are created automatically to enable inter-node communication within the cluster and communication from clients. Access is granted for clients by including values in the client_security_group_ids and/or client_cidr_blocks list variables.

To grant other kinds of access, such as SSH or XDCR, you may attach additional security groups using security_group_ids. Note that this change only takes effect on newly launched nodes/clusters, it will not affect existing nodes.

IAM Instance Profile

To support bootstrapping via the a rally node, the EC2 instances need the following security rights applied via an IAM role:

  • ec2:CreateTags
  • ec2:DescribeTags
  • ec2:DescribeInstances
  • autoscaling:DescribeAutoScalingGroups

By default, an IAM role and IAM instance profile are created for this purpose. If this is not an option in your environment for security reasons, you may supply your own profile via "iam_instance_profile".

Additional Initialization Script

You may provide some additional Bash script content to run on startup via additional_initialization_script. This runs after the node is fully initialized and joined to the cluster, and is useful for installing things like monitoring tools.

Auto Rebalancing

Auto rebalancing is enabled by default. This is useful for development and new clusters, but may not be desired for large production clusters. You may launch a new cluster with auto_rebalance enabled, then disable it and rerun terraform apply. This will disable auto rebalance for future node adds.

Upgrading Your Cluster using a Swap Rebalance

  1. Set auto_rebalance to false and set the new couchbase_version.
  2. Increase the size of your cluster by one node.
  3. Run terraform apply to apply the changes.
  4. Wait for the new node to join the cluster.
  5. In the Couchbase Web UI, remove one node with the same services. You should now have one node Pending Add and a matching node Pending Removal.
  6. Run a rebalance, and wait for it to complete.
  7. Terminate the EC2 instance you removed, which will cause the Auto Scaling Group to launch a replacement.
  8. Repeat steps 5-7 one node at a time until all nodes have been swapped.
  9. Use the AWS Console or CLI to remove the final leftover node from the Auto Scaling Group while decreasing the desired capacity.
  10. Terminate the leftover node

Support for Analytics Developer Preview

When deploying a Couchbase Server 5.5 cluster, you may wish to include the Analytics service. To do so, include the service on one of your MDS groups, set the RAM size for analytics, and supply an analytics_mpp:

services = {
    "1" = ["data"]
    "2" = ["index"]
    "3" = ["query"]
    "4" = ["analytics"]
}

cluster_ram_size = {
    data      = 8000
    index     = 4000
    fts       = 2000
    analytics = 8000
}

# For Analytics nodes with 4 CPU cores, this list has 4 unique values
analytics_mpp = [
    "0",
    "1",
    "2",
    "3",
]

Values for analytics_mpp should be unique and should not include path reserved characters. For optimum performance, you should have the same number of values in this list as the number of CPU cores on your analytics nodes.

About

Creates a complete Couchbase cluster in AWS, with support for Multidimensional Scaling

Resources

License

Stars

Watchers

Forks

Packages

No packages published