Add lb_logs_enabled var - defaults true

v8.1.0

add alb logs enabled var

Terraform 0.12.17 Upgrade

Merge pull request #16 from asicsdigital/casper-upgrade-12

Initial v12 upgrade

Add leave_on_terminate with default true for graceful leave

The consul default for leave_on_terminate is false; however running consul in docker means we need to set this to true: when the docker container exits, local data in the container is lost; therefore we need to send a graceful leave before exiting. Docker stop (and ECS Stop Task) send a TERM.

Note that without leave_on_terminate set to true, when ECS stops a consul task the cluster does still convert the node from failed to left via Forced Leave; however it would be preferred to send a graceful leave and not rely on the Forced leave.

Add lb_deregistration_delay variable

This release adds an optional variable lb_deregistration_delay; this allows to reduce the time required to cycle all ECS containers, for instance when cycling the underlying ECS cluster.

Note that by default this maintains the pre-existing default of 300 seconds.

Use Latest ACM Cert

Small Bugfix to use the latests issued ACM cert.

Improved Sidecar and Deployment behavior

This version of the module puts the requirement on the user of the
module to set instances to DRAINING in some fashion. In our case we are
using our sidecar container, which includes a healthcheck which set the
DRAINING State.

updates the sidecar container used.
This keeps this module inline with
FitnessKeeper/terraform-aws-consul-agents

In addition the old version of the module somewhat handled making sure the cluster would survive the instances being cycled under the ASG's by allowing you to span multiple ASG's *and *multiplying the desired task count by two, in this version the desired count is the correct count, so if you ask for a count of 3, you'll get three tasks, and we leave it to another facility to set the instances to draining. In this case, we provide all hooks to do this, but the parameter definitions still needs to be set correctly if you are using the default sidecar container image. The default setting is ["ecs-cluster"] which will correctly configure the sidecar to keep an eye on the instance life cycle state, and set the instance to DRAINING in ECS if the instance starts to terminate.

Adding Sidecar Container for configs and checks

The primary addition of this release is to add a sidecar container that can be used to add configs and health checks to the primary consul container, with the end goal of being able to update and use the upstream Consul container distributed by Hashicorp. The default sidecar is fitnesskeeper/consul-healthchecks:latest but this can be swapped out for another container image. In addition, the ECS Task definition has been updated to make it more compatible with the new versions of the terraform-aws-provider.

Adds Multi ECS cluster support

• Adds Multi ECS cluster support
• Adds datacenter_name parameter, an optional overide for datacenter name

Primarily this adds support to this module. This is a breaking change because the ecs_cluster_id parameter has become ecs_cluster_ids and is now a list as opposed to a string. The list supports two elements, and ignores any more then that. Passing a single element allows this module to work as it did in v4.0.0, passing in two ECS ID's creates two ECS services, and deploys the Consul Task across both. Was done to allow changes to instances in the autoscaling group, which might have effected Consul Cluster health to be done without rising the Consul cluster.

IAM Names Updated

This Release mostly updates how IAM resources are named. It is a breaking changed compared to the last release, but should provide more flexibility when creating multiple versions of a consul cluster across multiple regions in the same account.

Shorten ALB Names

This release is a fix for a bug uncovered by having a ALB name longer then 32 characters. Upgrading to this release will cause the cluster, and consul service to rebuild. - This will purge all data from your cluster - And is a breaking change.

IMPORTANT Backup your consul cluster to a snapshot and restore from snapshot after applying this version.