This can happen when CHECKDB is trying to read a page containing a row-over column (to construct a complete copy of a row in-memory). It follows a link to a text page and finds the page isn't allocated to the object it thought it was.
The error can happen for a number of reasons, all related to IO subsystem problems - for example, a lost write.
Can you run DBCC CHECKTABLE on the two tables in question and post the results (make sure to use the "WITH ALL_ERRORMSGS, NO_INFOMSGS" options on the DBCC commands).
Depending on the corruption, you may or may not lose data in the g2_pluginparametermap table, but my guess is that its likely.
How often do you run DBCC CHECKDB commands?
Did you run any IO soak/stress tests on the new array before using it?
SQL Server MVP, Managing Director, SQLskills.com