Attached file contains a sample of output. The output shows several errors are logged beginning with the 1000 node test, but I've seen that particular test work sometimes without the error being logged (when run along with the rest of the test class). If each test method is run separately, each test seems to pretty consistently run without logged errors.

Note that this problem does not result in a test failure.