If an RHTS test case fails on an uncommon architecture (ppc, s390, ...), or very intermittently, it would be mighty convenient if the machine could preserve its post-test state, and stay reserved awhile - say 12 hours. That way, an engineer has a fighting chance to reproduce the problem in situ, rather than hoping to reserve something similar, and hand-run the tests something similarly. I suspect the time saved by this may well make up for the reduced utilization of the machines, but if one disagrees, this behavior could be an option selected by rhts scripts.
You can now pass RESERVE_IF_FAIL=1 as a param to /distribution/reservesys and it will behave this way. *** This bug has been marked as a duplicate of bug 623933 ***