CHANGELOG.md

2.17.0 (2022-05-31)

New features

• Support runtime_platform option

2.16.0 (2022-05-02)

New features

• Support ephemeral_storage option in ECS scheduler

Bug fixes

• Fix keyword arguments for Ruby 3.0

Changes

• Drop support for Ruby 2.5 (EOL on 2021-04-05)

2.15.1 (2021-08-06)

Bug fixes

• Fix keyword arguments for Ruby 3.0

2.15.0 (2020-11-02)

New features

• Support protocol_version and matcher option of ALB target groups

2.14.0 (2020-05-20)

New features

• Support tags for task definition and propagate them to ECS tasks
  • Now all created ECS services and launched ECS tasks have propagate_tags=TASK_DEFINITION parameter.
• Support repository_credentials

2.13.0 (2020-01-10)

New features

• Support capacity provider strategy

Bug fixes

• Do not try to update assign_public_ip when it is not changed

2.12.0 (2019-09-09)

New features

• Support more overrides options for hako oneshot
  • --app-cpu, --app-memory and --app-memory-reservation are added

Bug fixes

• Show --health-* options in dry-run

2.11.1 (2019-05-17)

Bug fixes

• Fix comparison of system_controls parameter

2.11.0 (2019-05-17)

New features

• Support system_controls parameter in container definition

2.10.0 (2019-05-09)

New features

• Support depends_on parameter in container definition

2.9.2 (2019-04-10)

Bug fixes

• Change threshold for detecting a deployment failure
  • See this commit for details

2.9.1 (2019-04-04)

Bug fixes

• Skip creating load balancer on dry-run
  • Regression in v2.9.0

2.9.0 (2019-04-03)

New features

• Add remove_starting method to Hako::Script which is called when hako remove starts
• Support sharing load balancers
  • When elb_v2 field has load_balancer_name but doesn't have target_group_name, hako will manage only the target group and doesn't touch the load balancer.
  • This is useful when one load balancer is shared with multiple target groups and some of them is deployed by hako.
  • See examples/hello-shared-alb.jsonnet

Changes

• When load_balancer_name is specified and target_group_name is not, the default target_group_name is changed from load_balancer_name to hako-#{app_id}.
  • If you use load_balancer_name field only, you must specify target_group_name field too.

2.8.0 (2019-03-17)

New features

• Add deploy_failed method to Hako::Script which is called when hako deploy fails
• Add .app.tag field to definition to specify the default value of app container's tag
  • The default value was fixed to "latest", but you can now specify custom default value.

2.7.0 (2019-03-15)

New features

• Support entry_point parameter
• Support ECS Service Discovery
  • See examples/hello-service-discovery.jsonnet

2.6.2 (2018-12-19)

Bug fixes

• Set platform_version correctly

2.6.1 (2018-12-17)

Bug fixes

• Avoid updating service when platform_version is not specified
• Keep essential parameter in hako oneshot

Improvements

• Show more information about tasks in hako status

2.6.0 (2018-12-13)

New features

• Support essential parameter for each container
  • The default value remains true

Bug fixes

• Stop trying to deregister non-existent scalable target on hako remove
• Update service when health_check_grace_period_seconds changes

2.5.1 (2018-11-29)

Bug fixes

• Check secrets existence in dry-run mode

2.5.0 (2018-11-22)

New features

• Add load_balancer_name and target_group_name option to skip creating ELB
• Support ECS secrets

2.4.0 (2018-11-13)

New features

• Support readonly_root_filesystem parameter
• Support docker_security_options parameter
• Support ssl_policy option in elb_v2

Bug fixes

• Skip updating desired_count before removing service using daemon scheduling strategy

2.3.1 (2018-09-26)

Changes

• Change show-definition output from YAML to JSON
  • The show-definition output is still parsable as YAML

Bug fixes

• Set deployment_configuration to nil when absent

2.3.0 (2018-08-30)

New features

• Support health_check parameter
• Support shared_memory_size parameter
• Support tmpfs parameter
• Support docker_volume_configuration parameter
• Support target tracking scaling policy

Changes

• Run containers referenced by volumes_from on hako oneshot
• Rename additional_containers parameter to sidecars
  • additional_containers is still supported for compatibility

Bug fixes

• Show --volumes-from in dry-run output
• Fix --init not being shown in dry-run output

2.2.0 (2018-06-29)

New features

• Add support for scheduling_strategy on service
• Change existing ELB's subnets when different from definition

Bug fixes

• Take task-level cpu/memory into account on scale out
• Show --memory-reservation in the dry-run output

2.1.0 (2018-04-18)

New features

• Support Network Load Balancer
  • See examples/hello-internal-nlb.jsonnet
• Support health_check_grace_period_seconds
  • See examples/hello-lb-v2.jsonnet

2.0.4 (2018-02-26)

Bug fixes

• Pass AWS region of ECS scheduler to other AWS clients (CloudWatch, ApplicationAutoScaling)
• Eliminate --memory parameter from dry-run output if it's not given
• Give missing unit to the value of --memory parameter in dry-run output
• Take memory_reservation into account when calculating required memory

2.0.3 (2018-02-16)

Bug fixes

• create_aws_cloud_watch_logs_log_group script: Skip creating CloudWatch log group on dry-run

2.0.2 (2017-12-19)

Bug fixes

• Skip expanding variables in remove and stop

2.0.1 (2017-12-14)

Bug fixes

• Fix compatibility between Jsonnet and YAML when --dry-run is given

2.0.0 (2017-12-13)

New features

• Support Jsonnet as the definition file format
  • See docs/jsonnet.md
  • YAML definitions continues to be supported by hako at least v2.x series. Currently I don't have a plan which will be the default format.

1.9.0 (2017-12-06)

New features

• Add support for awsvpc network mode
• Add support for Fargate
  • See examples/hello-fargate.jsonnet and examples/hello-fargate-batch.jsonnet

1.8.4 (2017-11-24)

Improvements

• Support linux_parameters option
• S3 and SNS client now uses the same region with ECS

1.8.3 (2017-09-25)

Improvements

• Support extra_hosts option

1.8.2 (2017-09-15)

Bug fixes

• Retry DeleteTargetGroup in hako remove
  • Deleting a load balancer may take several seconds

Improvements

• Add target_group_attributes option to elb_v2

1.8.1 (2017-09-15)

Improvements

• Add container_name and container_port option to elb and elb_v2
  • See examples/hello-nofront.jsonnet

1.8.0 (2017-09-15)

Changes

• Migrate to aws-sdk v3

1.7.0 (2017-08-29)

New features

• Add experimental autoscaling_topic_for_oneshot option to ECS scheduler
  • It publishes scale-out request to SNS topic.
  • Administrators is expected to receive SNS event and initiate scale-out.

1.6.2 (2017-07-11)

Bug fixes

• Exclude unusable instances when checking remaining capacity

1.6.1 (2017-06-26)

Changes

• Output cluster information in --no-wait mode

1.6.0 (2017-06-23)

New features

• Add experimental option --no-wait to oneshot
  • hako oneshot --no-wait runs Docker container in background and return an identifier depending on scheduler.
  • In ECS scheduler, it will output the task's ARN.

1.5.2 (2017-06-08)

Bug fixes

• Retry RegisterTaskDefinition when "too many concurrent attempts" error occurs

1.5.1 (2017-06-06)

Bug fixes

• Fix error in dry-run mode when ALB isn't created yet

1.5.0 (2017-06-05)

New features

• Support load_balancer_attributes option in elb_v2

1.4.0 (2017-05-25)

New features

• Support ulimits option

1.3.3 (2017-05-23)

Bug fixes

• Fix error of autoscaling for oneshot when container instances are empty
• Fix error when new task definition is registered
  • Regression in v1.3.2

1.3.2 (2017-05-23)

Bug fixes

• Pass placement_configurations in oneshot mode
• Symbolize port_mapping keys to compare definitions correctly

Improvements

• Reduce the number of DescribeTaskDefinition calls

1.3.1 (2017-05-16)

Bug fixes

• Retry DescribeAutoScalingGroups when rate limited

1.3.0 (2017-05-15)

New features

• Add oneshot_notification_prefix option
  • This is experimental , so might be reverted in near version.
  • This option enables S3 polling instead of ECS polling.

1.2.1 (2017-05-11)

Bug fixes

• Retry DescribeTasks when rate limited

1.2.0 (2017-04-14)

Changes

• Fail deployment when some tasks are stopped during deployment
  • It should prevent infinite loop when the new revision always fails to start

1.1.0 (2017-03-09)

New features

• Add script hooks to rollback
  • Script#rollback_starting
    • Similar to deploy_starting, but without containers
  • Script#rollback_started
    • Current running image tag and target image tag are passed
  • Script#rollback_finished
    • Similar to deploy_finished, but without containers

1.0.1 (2017-03-07)

Bug fixes

• Fix default value of @volumes
  • When mount_points is specified but no volumes is given, hako was raising unfriendly errors

1.0.0 (2017-03-06)

Incompatibilities

• Raise error when env value is not String (#31)