Adding some options multiple times to the command-line ends up with a list of values in config

[jkovacs-hwx]

E.g. using

hms-mirror  \
--external-warehouse-directory /warehouse/tablespace/external/hive \
... \
--external-warehouse-directory /warehouse/tablespace/external/hive \
...

ends up with a config value:

externalDirectory: "/warehouse/tablespace/external/hive,/warehouse/tablespace/external/hive"

...which could cause issues at runtime.

Command should either fail or should use only the last defined value for configs that should have single values.

[dstreev]

This looks like a bug in the commons-cli library where the values are combined even though the option has been configured to accept only 1 argument.

[jkovacs-hwx]

I wasn't sure how these are picked up but tested and found that unfortunately used as a single value:
Using:

  --external-warehouse-directory /data/hive/external --warehouse-directory /data/hive/managed \
  --external-warehouse-directory /data/hive/external --warehouse-directory /data/hive/managed

Got in session-config.yaml:

  warehouse:
    source: "GLOBAL"
    externalDirectory: "/data/hive/external,/data/hive/external"
    managedDirectory: "/data/hive/managed,/data/hive/managed"

and in RIGHT_execute.sql:

-- Alter Database Location
ALTER DATABASE mydb SET LOCATION "hdfs://cdpprod/data/hive/external,/data/hive/external/hive";
-- Alter Database Managed Location
ALTER DATABASE mydb SET MANAGEDLOCATION "hdfs://cdpprod/data/hive/managed,/data/hive/managed/mydb";

[dstreev]

Yes, agreed. In my tests it looks like the commons-cli library is picking up both values and combining them before I intercept them for evaluation.