It appears that MongoDB has very slow writes (108 ms / record locally) with the FsyncSafe WriteConcern. I suppose this is new in the 2.4 series of MongoDB, as it wasn't the case earlier on. This caused a congestion in the processing pipeline which eventually leads to an out-of-memory error. I am now relying on MongoDBs claim that records will eventually get written from memory to disk. Additionally, there's now a concrete number displayed next to the progress bar.
…t check solr for hubId if we are dealing with indexItems.
…mers expecing the hubId to be available straight away get the right response.