- Fix security validation check
- Vault connection certificates are obtained dynamically
- KMS Utils dependency remove from Stratio Spark Dispatcher and History Server
- Possibility of specifying the name of the certificate within a directory with multiple certificates
- Add support to multiples CAs
- Added performance tests PNF
- Fix logs to be same as rest of Stratio platform
- Add messages for errors in Vault
- Fixed Supervise mode
- Separate stderr and stdout in dispatcher
- Fix history server stderr/stdout. Now is possible to set log level through SPARK_LOG_LEVEL
- Removed mesos security from History Server and unified environment variable VAULT_HOSTS
- Secret folder path configurable
- Changed log format according to Stratio standards
- Mesos Role no longer obtain mesos pricipal and mesos secret from vault in Spark jobs
- Added Tenant Name variable being able to have Spark Dispatcher with different name
- Enable Debug mode
- Support Dynamic allocation with Calico (External shuffle working in HOST mode)
- Unify Vault variables
- Secret Broadcast variables (Experimental)
- Added mesos constraints management to spark driver
- Added a secure way to retrieve user and passwords information from vault
- History Server could read from HDFS in HA using environment variables
- Added validation to curl parameters
- Removed Mesos secret and principal from curls
- Added configurable HDFS timeout
- Connection to Elastic with TLS
- Connection to Postgres with TLS, unified in datastore identity
- Removed Kafka identity, unified in datastore identity
- Removed script connection to Postgres
- Spark Dispatcher retrieves Mesos Principal and Secret from Vault
- Fix History Server env vars
- Dynamic Authentication for History Server
- SDN compatibility and isolation for History Server
- Refactor vault variables
- Spark-vault interactions
- SDN compatibility and isolation
- Kerberized access to hdfs
- Postgress TLS connection
- Dynamic Authentication
- Stratio Mesos security compatibility
- Initial Stratio Version of Spark
- Forked from Apache Spark 2.1.0