A Constraint must be added to a table before the table is loaded via Constraints.add(HTableDescriptor, Class...) or Constraints.add(HTableDescriptor, org.apache.hadoop.hbase.util.Pair...) (if you want to add a configuration with the Constraint). Constraints will be run in the order that they are added. Further, a Constraint will be configured before it is run (on load).

See Constraints.enableConstraint(HTableDescriptor, Class) and Constraints.disableConstraint(HTableDescriptor, Class) for enabling/disabling of a given Constraint after it has been added.

If a Put is invalid, the Constraint should throw some sort of ConstraintException, indicating that the Put has failed. When this exception is thrown, not further retries of the Put are attempted nor are any other Constraints attempted (the Put is clearly not valid). Therefore, there are performance implications in the order in which Constraints are specified.

If a Constraint fails to fail the Put via a ConstraintException, but instead throws a RuntimeException, the entire constraint processing mechanism (ConstraintProcessor) will be unloaded from the table. This ensures that the region server is still functional, but no more Puts will be checked via Constraints.

Further, Constraints should probably not be used to enforce cross-table references as it will cause tremendous write slowdowns, but it is possible.

NOTE: Implementing classes must have a nullary (no-args) constructor