Stress Tests induce incorrect value

[Peter Beaman]

Observed while running persistit-stress-multi-config2 on an ec2 instance:

Starting Stress10Suite for page size 4,096
deleted /mnt/persistit_tests/persistit
deleted /mnt/persistit_tests/persistit_journal.000000000000
deleted /mnt/persistit_tests/persistit_201211100123.log
Stress10Suite at 60 seconds: live= 12 ended= 0 stopped = 0, failed= 0 totalwork= 9,758,916 intervalwork= 9,758,916 workrate= 160,456
Stress10Suite at 120 seconds: live= 12 ended= 0 stopped = 0, failed= 0 totalwork= 15,563,439 intervalwork= 5,804,523 workrate= 129,339
Stress10Suite at 180 seconds: live= 12 ended= 0 stopped = 0, failed= 0 totalwork= 16,315,113 intervalwork= 751,674 workrate= 90,173

Stress10[Thread-627]: FAILED: Value is size 0, should be 2370 key={38833}

Stress10[Thread-624]: FAILED: Value is size 0, should be 463 key={20663}
Stress10Suite at 221 seconds: live= 10 ended= 2 stopped = 0, failed= 2 totalwork= 16,628,862 intervalwork= 313,749 workrate= 75,131
Stress10Suite at 222 seconds: live= 0 ended= 12 stopped = 10, failed= 2 totalwork= 16,628,909 intervalwork= 47 workrate= 74,790

---Result Stress10Suite: FAILED work=16,628,909 time=222 rate=74,904 ---

[Peter Beaman (pbeaman)]

It is possible that the test itself is wrong.

[Peter Beaman (pbeaman)]

Observed in branch lp:~pbeaman/akiban-persistit/find-deadlock3 rev 391

[Peter Beaman (pbeaman)]

This bug resulted from faulty code added to handle RebalanceExceptions and was never found in trunk. It has now been resolved.