It is possible that the JDBC driver to iSeries (DB2/400) returns untranslated EBDIC characters instead of characters in Unicode strings. The symptoms of this issue are fields read by CloverETL but not viewable.

To solve this issue, set the translate binary property of the jdbc driver to true.

See http://www-03.ibm.com/systems/power/software/i/toolbox/faq/jdbc.html#faqB8

Or you can modify the CCSID with CHGPF command. See https://www.ibm.com/support/knowledgecenter/ssw_ibm_i_71/cl/chgpf.htm