Use batch processing methods in Hierarchy Maintenance when looping through database records

[nbekirov]

The problem

I'm migrating from system where my model already has aparent_id column so following the documentation when I tried running Tag.rebuild! to have my tag_hierarchies table truncated and rebuilt I got an error.

Initially this resulted in Java::JavaLang::OutOfMemoryError: GC overhead limit exceeded (with the default JVM heap size of 512MB).

A solution

Expanding the memory via JRUBY_OPTS=-J-Xmx32G did the trick and rebuild! passed successfully but with ever expanding heap.

You can take a look of this jvisualvm screenshots made before the proposed change:

and after:

This even allows my 512MB setup to succeed:

Context

• My test database is populated with ~49k nodes that results in ~240k rows hierarchy and is roughly 10 times smaller than production.
• The setup:
  • macOS Sierra 10.12.3, 3.5 GHz Intel Core i5, 8 GB 1600 MHz DDR3
  • jruby-9.1.2.0
  • PostgreSQL 9.6.2

[mceachen]

This was so nice I pushed a new version. Thanks for v6.5.0!