When starting with a huge number of threads, timestamp_abort hangs and eventually aborts in the logging code. The code cannot find a free log slot for 10 seconds and then dumps out all the log slot information.

The large increase in the maximum number of threads in the test is from WT-3995. This failure on kodkod at http://build.wiredtiger.com:8080/job/wiredtiger-test-recovery-stress/52365/console had this configuration:

test_timestamp_abort -h WT_TEST.timestamp-abort -T 195 -t 10