Currently much of the logging has been handle using System Out. This should be changed to use a logger, and Log Level should be added to the docker file as a configurable environment variable.

An extractor will need an element collection of new embeddable POJO to know how to make subsequent database lookups per result row of stream. The additional lookups should use a single database connection.

New POJO will require SQL template for new lookup query and a strategy for handling the results. The template should use the stream row as its context. The initial strategy will be to order results by column and aggregate by another column as a string. This string will be added to the stream row.

Refer to https://github.com/TAMULib/mod-data-migration#marc-bibliographic-migration for bibliographic migration.

Here is an example of additional lookup and strategy, https://github.com/TAMULib/mod-data-migration/blob/master/src/main/java/org/folio/rest/migration/BibMigration.java#L421

2019-11-18 10:05:42.040 INFO 11272 --- [nio-9002-exec-3] o.f.rest.controller.ExtractorController : Running extractor ed3fa3de-4341-4960-bf7e-552ffd2ad04f
2019-11-18 10:07:18.512 ERROR 11272 --- [task-executor-1] tractorStreamingResponseArgumentResolver : java.net.SocketTimeoutException
2019-11-18 10:07:18.520 ERROR 11272 --- [task-executor-1] tractorStreamingResponseArgumentResolver : java.io.IOException: The current thread was interrupted
2019-11-18 10:08:18.530 ERROR 11272 --- [task-executor-1] tractorStreamingResponseArgumentResolver : java.net.SocketTimeoutException

Last exception it was 15 minutes. This time was not.
May be back pressure causing this issue.