First Astarte v1.2.1 release candidate.

Changes since 1.2.1-alpha.0

Added

• New environment variables to control how clustering work, needed on AppEngine and DUP.
  • CLUSTERING_STRATEGY. Its possible values are:
    • none (default): the service will not look for other nodes/services.
    • docker-compose: this is meant to work in our docker-compose environment and no additional configuration is needed.
    • kubernetes: other nodes/services are found thanks to kubernetes DNS, and the other variables below are relevant.
  • CLUSTERING_KUBERNETES_NAMESPACE. It states under which namespace the Astarte instance has been deployed. Defaults to astarte.
  • DATA_UPDATER_PLANT_CLUSTERING_KUBERNETES_SELECTOR. The Endpoint label to query to get other data updater plant instances. Defaults to app=astarte-data-updater-plant.
  • VERNEMQ_CLUSTERING_KUBERNETES_SELECTOR. The Pod label to use to query Kubernetes to find VerneMQ instances. Defaults to app=astarte-vernemq.
  • VERNEMQ_CLUSTERING_KUBERNETES_SERVICE_NAME. The Service name to use to query Kubernetes to find VerneMQ instances. Defaults to astarte-vernemq.

Changed

• Rework RPC between AppEngine and DUP using Erlang's native clustering and message-passing instead of AMQP queues: #1186. Fix #699.
• Rework RPC between AppEngine and VerneMQ using Erlang's native clustering and message-passing instead of AMQP queues: #1194.
• Rework RPC between DUP and VerneMQ using Erlang's native clustering and message-passing instead of AMQP queues: #1197.
• Changed the database driver from CQEx (unmantained) to (E)xandra
• [astarte_trigger_engine] avoid exposing unknown_fields in mustache templates
• [astarte_trigger_engine] properly handle incoming introspection events

Fixed

• avoid leaving dangling device deletion entries
• [astarte_appengine_api] fix a crash on invalid object update values
• [astarte_appengine_api] fix a crash with empty result and disjoint_tables format for object aggregates
• [astarte_appengine_api] Create needed AMQP exchanges instead of crashing
• [astarte_appengine_api] Handle unset of properties which don't allow being unset
• [astarte_data_updater_plant] fix a crash when performing updates of deleted values
• [astarte_data_updater_plant] Fix a crash while handling device introspection
• [astarte_data_updater_plant] Fix DataUpdater GenServer timeout handling that prevented inactive processes to shut down automatically.
• [astarte_data_updater_plant] Handle unset of properties which don't allow being unset
• [astarte_data_updater_plant] properly discard heartbeat messages when discarding messages
• [astarte_data_updater_plant] Properly reconnect to RabbitMQ in case of disconnection
• [astarte_data_updater_plant] Some queries had the astarte_instance_id applied twice
• [astarte_realm_management] avoid crashing on interface list
• [astarte_realm_management] avoid crash when deleting devices with invalid introspection
• [astarte_realm_management] ensure devices are cleaned up after being deleted