Datamapper DataMapper

We use multi-repository access throughout our DM code. One practical upshot of this is that batch scripts that stripe through all of the repositories end up holding open (due to DO connection pooling) huge numbers of surplus connections to the DB. In order to combat this, we've wrapped batch tasks in the following genre of code...

    scavenger_interval = DataObjects::Pooling.scavenger_interval
    DataObjects::Pooling.scavenger_interval = 2
    
    ...some code here...
    
    DataObjects::Pooling.scavenger_interval = scavenger_interval

We're seeing this blow up for certain kinds of operations. The error raised is actually a SystemExit...

    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:60:in `scavenger'
    from /usr/lib/ruby/1.8/set.rb:194:in `each'
    from /usr/lib/ruby/1.8/set.rb:194:in `each_key'
    from /usr/lib/ruby/1.8/set.rb:194:in `each'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:57:in `scavenger'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:56:in `synchronize'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:56:in `scavenger'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:49:in `initialize'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:49:in `new'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:49:in `scavenger'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:90:in `append_pool'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:154:in `initialize'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:119:in `new'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:119:in `new'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:119:in `synchronize'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/pooling.rb:119:in `new'
    from /home/vjuser/.bundle/ruby/1.8/gems/data_objects-0.10.1/lib/data_objects/connection.rb:63:in `new'
    from /home/vjuser/.bundle/ruby/1.8/gems/dm-core-0.10.2/lib/dm-core/adapters/data_objects_adapter.rb:233:in `open_connection'
    from /home/vjuser/.bundle/ruby/1.8/gems/dm-core-0.10.2/lib/dm-core/transaction.rb:412:in `open_connection'
    from /home/vjuser/.bundle/ruby/1.8/gems/dm-core-0.10.2/lib/dm-core/adapters/data_objects_adapter.rb:269:in `with_connection'
    from /home/vjuser/.bundle/ruby/1.8/gems/dm-core-0.10.2/lib/dm-core/adapters/data_objects_adapter.rb:136:in `read'
    from /home/vjuser/.bundle/ruby/1.8/gems/dm-core-0.10.2/lib/dm-core/repository.rb:145:in `read'
    from /home/vjuser/.bundle/ruby/1.8/gems/dm-core-0.10.2/lib/dm-core/collection.rb:1111:in `lazy_load'
    from /home/vjuser/.bundle/ruby/1.8/gems/extlib-0.9.15/lib/extlib/lazy_array.rb:411:in `each'
    from /home/vjuser/.bundle/ruby/1.8/gems/dm-core-0.10.2/lib/dm-core/collection.rb:506:in `each'

I note this is occurring on/in a Mutex.synchronize so I assume we're dealing with a hairy threading problem.