Before a pipeline can be run, the eHive database must be created and populated. This is done using eHive’s init_pipeline.pl script.
Two parameters are required by
- The fully qualified Perl classname of the PipeConfig to be initialised.
- The database to be initialised, typically passed as a url using
-pipeline_url[*] . It can also be passed in using
-password, or it can be specified in the
init_pipeline.pl MyProject::PipeConfig::ExamplePipeline_conf -pipeline_url mysql://user:firstname.lastname@example.org:4567/my_example_pipeline_db
When this script is run, it performs the following operations:
- It creates a new database with the name given.
init_pipeline.plis called with
-hive_force_init 1it will also remove any database with the same name that may exist - e.g. with a
DROP DATABASE IF EXISTSin MySQL.
- It creates the table structure, along with foreign keys, views, and stored procedures (if applicable).
init_pipeline.pl successfully completes, it will print a series of useful
commands for working with the now-instantiated pipeline.
There are a few common sources of problems that may be encountered when running
- During pipeline initialisation, all of the Runnables listed in the given PipeConfig will be compiled. If the Runnables cannot be located by Perl, Python, or Java then initialisation will fail. Errors in a Runnable that prevent compilation, such as syntax errors, will also prevent the pipeline from initialising.
- If the given database already exists, initialisation will fail unless the
-hive_force_init 1option has been given. This protects existing pipeline instances from accidental deletion.
- If the user does not have sufficient permissions (CREATE, CREATE ROUTINE, INSERT, and ALTER) in the given database, initialisation will fail.
|[*]||Note the name |