Autoscaler/autoscaler v2.0.0

on GitHub

Version Number

2.0.0-914

Breaking Changes

709003: RabbitMQ disk space will now be monitored by default.

The following environment variables have been renamed:

• CAF_AUTOSCALER_RABBITMQ_MEMORY_QUERY_FREQ renamed to CAF_AUTOSCALER_RABBITMQ_RESOURCE_QUERY_FREQ
• CAF_AUTOSCALER_MESSAGING_RESOURCE_LIMIT_STAGE_1 renamed to CAF_AUTOSCALER_MESSAGING_MEMORY_USED_PERCENT_LIMIT_STAGE_1
• CAF_AUTOSCALER_MESSAGING_RESOURCE_LIMIT_STAGE_2 renamed to CAF_AUTOSCALER_MESSAGING_MEMORY_USED_PERCENT_LIMIT_STAGE_2
• CAF_AUTOSCALER_MESSAGING_RESOURCE_LIMIT_STAGE_3 renamed to CAF_AUTOSCALER_MESSAGING_MEMORY_USED_PERCENT_LIMIT_STAGE_3
• CAF_AUTOSCALER_ALERT_DISPATCH_THRESHOLD renamed to CAF_AUTOSCALER_MEMORY_USED_PERCENT_ALERT_DISPATCH_THRESHOLD

The following environment variables have been added:

• CAF_AUTOSCALER_DISK_FREE_MB_ALERT_DISPATCH_THRESHOLD

  • The amount of disk space (MB) remaining on the messaging platform that will trigger the Autoscaler to dispatch alerts indicating that the messaging platform is running out of disk space.
  • If this is not set then a message will be dispatched from the CAF_AUTOSCALER_MESSAGING_DISK_FREE_MB_LIMIT_STAGE_1 value, or 400 if CAF_AUTOSCALER_MESSAGING_DISK_FREE_MB_LIMIT_STAGE_1 is not set.

• CAF_AUTOSCALER_MESSAGING_DISK_FREE_MB_LIMIT_STAGE_1

  • The amount of disk space (MB) remaining on the messaging platform that will trigger the Autoscaler to take stage 1 action.
  • Stage 1 action will involve shutting down any services with a shutdown priority of less than or equal to CAF_AUTOSCALER_MESSAGING_STAGE_1_SHUTDOWN_THRESHOLD or 1 if the environment variable is not set.
  • Default value is 400

• CAF_AUTOSCALER_MESSAGING_DISK_FREE_MB_LIMIT_STAGE_2

  • The amount of disk space (MB) remaining on the messaging platform that will trigger the Autoscaler to take stage 2 action.
  • Stage 2 action will involve shutting down any services with a shutdown priority of less than or equal to CAF_AUTOSCALER_MESSAGING_STAGE_2_SHUTDOWN_THRESHOLD or 3 if the environment variable is not set.
  • Default value is 200

• CAF_AUTOSCALER_MESSAGING_DISK_FREE_MB_LIMIT_STAGE_3

  • The amount of disk space (MB) remaining on the messaging platform that will trigger the Autoscaler to take stage 3 action.
  • Stage 3 action will involve shutting down any services with a shutdown priority of less than or equal to CAF_AUTOSCALER_MESSAGING_STAGE_3_SHUTDOWN_THRESHOLD or 5 if the environment variable is not set.
  • Default value is 100

New Features

693063: Staging queues can now be considered when making a scaling decision.

• An optional environment variable named CAF_AUTOSCALER_STAGING_QUEUE_INDICATOR can now be provided.
• If this is provided, queue names that follow the naming pattern "^" + scalingTarget + stagingQueueIndicator + ".+#" will be considered
  staging queues and the statistics for these staging queues will be taken into account alongside the statistics for the target queue
  (scalingTarget) when making a scaling decision.

Bug Fixes

709003: An issue preventing memory overload alerts from being dispatched when the messaging platform was experiencing high memory usage has been fixed.

709003: An issue preventing instances from being scaled down in Kubernetes when the messaging platform was experiencing high memory usage has been fixed.

• This was caused by the Autoscaler attempting to read the autoscale.shutdownPriority label instead of the autoscale.shutdownpriority label.

Known Issues

• None