[ https://issues.jboss.org/browse/DROOLS-1326?page=com.atlassian.jira.plugi... ] Jiri Locker commented on DROOLS-1326: ------------------------------------- So what I did is that I removed the 1st solver, which only does a part of CH. Then I took the 2nd solver running with FAST_ASSERT, I let it do the whole CH (740 steps total), and I limited its LS to 1233 steps, which should lead to the same state as the original 1st and 2nd solvers without any modifications. Then I left the 3rd solver with NON_INTRUSIVE_FULL_ASSERT to solve for 30 minutes, supposedly starting from the same state, but it didn't fail. The output with the described modifications looks like this: {noformat} 2016-10-14 14:32:48,296 [main] INFO Opened: data/nurserostering/unsolved/long_hint01.xml 2016-10-14 14:32:49,902 [main] INFO Solving started: time spent (177), best score (-740init/0hard/-1580soft), environment mode (FAST_ASSERT), random (JDK with seed 0). 2016-10-14 14:32:54,252 [main] INFO Construction Heuristic phase (0) ended: time spent (4527), best score (0hard/-180soft), score calculation speed (17355/sec), step total (740). 2016-10-14 14:36:46,131 [main] INFO Local Search phase (1) ended: time spent (236406), best score (0hard/-62soft), score calculation speed (8812/sec), step total (1233). 2016-10-14 14:36:46,132 [main] INFO Solving ended: time spent (236407), best score (0hard/-62soft), score calculation speed (8962/sec), phase total (2), environment mode (FAST_ASSERT). 2016-10-14 14:36:46,630 [main] INFO Solving started: time spent (153), best score (0hard/-62soft), environment mode (NON_INTRUSIVE_FULL_ASSERT), random (JDK with seed 0). 2016-10-14 14:36:46,631 [main] INFO Construction Heuristic phase (0) ended: time spent (154), best score (0hard/-62soft), score calculation speed (1000/sec), step total (0). {noformat} -- This message was sent by Atlassian JIRA (v6.4.11#64026)