soda-fountain is Socrata's HTTP REST server for serving Socrata Open Data API (SODA) version 2 API requests. It talks to two other REST microservices, the data-coordinator, and the query-coordinator, to manage both CRUD operations on datasets as well as queries against datasets.
- soda-fountain-lib -- most of the code lives here, including routes, request handling, etc.
- soda-fountain-jetty - a Main for running within embedded Jetty
- soda-fountain-war - stuff for running in an application container
- Zookeeper
- PostgreSQL (stores metadata about datasets)
- Query Coordinator (registered through Zookeeper)
- Data Coordinator (registered through Zookeeper)
For fast dev cycles:
- Start SBT shell
reStart --- -Dconfig.file=/etc/soda2.conf
This forks a separate JVM, killing any previously forked JVM. For really fast compile and auto run, do ~reStart. Much faster than building an assembly!
To build an assembly and run as a separate process:
bin/start_soda_fountain.sh
sbt "soda-fountain/test"
For soda-fountain-lib, log output goes to sbt-test.log. Logging is controlled via soda-fountain-lib/src/test/resources/log4j.properties.
For test coverage reports, do sbt coverage test. XML and HTML coverage reports are generated. Unfortunately due to a bug in scalac, HTML line by line highlighting is broken.
Using sbt:
sbt -Dconfig.file=/etc/soda2.conf "soda-fountain-jetty/run-main com.socrata.soda.server.MigrateSchema [command] [numberOfChanges]"
For example:
sbt -Dconfig.file=/etc/soda2.conf "soda-fountain-jetty/run-main com.socrata.soda.server.MigrateSchema migrate"
To build and run migrations from command line:
bin/run_migrations.sh
- migrate - apply all migrations to the database
- undo - rollback the latest change, or [numberOfChanges] if specified
- redo - runs undo then migrate in one command
To publish the external library:
- update the version.sbt file, create a PR and merge to main
- build the soda-fountain-branch job: check PUBLISH and enter the SHA from the version update commit in PUBLISH_SHA