-
Notifications
You must be signed in to change notification settings - Fork 8
4. Advanced Tips for Deployment
For storing the data, it is advisable to use Docker volumes. For that, first create a volume using
docker volume create adata
When running the container, mount the volume to the destination /adampro/data
where all the data is stored.
docker run ... --mount source=adata,destination=/adampro/data ...
With that you may create your entities and fill these with data. When deleting the container, the volume is kept and can be attached to a new container.
For accessing the underlying storage engines directly, expose the necessary ports. Note that not all ports might be available (depending on the container you are using).
-
5890 for the ADAMpro grpc endpoint
-
9099 for the ADAMpro Web UI
-
4040 for the Apache Spark UI
-
50010 50020 50070 50075 50090 for Apache Hadoop (HDFS)
-
1988 for Apache Hadoop (MapReduce)
-
8030 8031 8032 8033 8040 8042 8088 for Apache Hadoop (Yarn)
-
5432 for PostgreSQL
-
8983 for Apache Solr
-
9042 for Apache Cassandra
-
10088 for the Spark Notebook
-
19999 for netdata
Adjust the environment variables ADAMPRO_EXECUTOR_MEMORY
and ADAMPRO_DRIVER_MEMORY
(for the default container and the hdfs container, set to 2g
) or ADAMPRO_MEMORY
(for the selfcontained container, set to 2g
), respectively, denoting how much memory is allocated to ADAMpro and Apache Spark, and ADAMPRO_MASTER
(set to local[4]
), which sets the number of threads in the local deployment. Both parameters have very minimal default values.
In the selfcontained
container, you may choose which storage engines to start. You can do this by adjusting the following environmental variables:
-
ADAMPRO_START_POSTGRES
(default: true) -
ADAMPRO_START_CASSANDRA
(default: true) -
ADAMPRO_START_SOLR
(default: true) -
ADAMPRO_START_WEBUI
(default: true) -
ADAMPRO_START_NOTEBOOK
(default: false) -
ADAMPRO_START_NETDATA
(default: true)
Certain containers are configured to have a graceful shutdown. Use
docker stop --time 60 adampro
and give the container around 1 minute time to gracefully shutdown all services.