As per parent task, it's possible that the origin computation urls was mixed up.
Resulting in the same origin for different dumps (googlecode hosted googlecode and eclispselabs dumps in different arborescence tree but that was not accounted for when creating the actual origin's url).
This needs to be investigated and fixed if it's actually true (at least one occurrence is referenced in the parent so that tends to be :).
Note:
This would also explain an error so far unexplained 'No revision found' on the latest svn loader rescheduling.