The test failure after merge of WT-2692 made me wonder why the pull request tester didn't flag the error. When I looked at the output I see that it's not running the python test suite:
+ env LD_LIBRARY_PATH=.libs python ../test/suite/run.py -j 4 -v 4 -j 12 Usage: $ cd build_posix $ python ../test/suite/run.py [ options ] [ tests ] Options: -C file | --configcreate file create a config file for controlling tests -c file | --config file use a config file for controlling tests -D dir | --dir dir use dir rather than WT_TEST. dir is removed/recreated as a first step. -d | --debug run with 'pdb', the python debugger -g | --gdb all subprocesses (like calls to wt) use gdb -h | --help show this message -j N | --parallel N run all tests in parallel using N processes -l | --long run the entire test suite -p | --preserve preserve output files in WT_TEST/<testname> -t | --timestamp name WT_TEST according to timestamp -v N | --verbose N set verboseness to N (0<=N<=3, default=1) Tests: may be a file name in test/suite: (e.g. test_base01.py) may be a subsuite name (e.g. 'base' runs test_base*.py) When -C or -c are present, there may not be any tests named. + make clean
There are two problems that require fixes:
1) Why isn't the command correct
2) Why isn't the failed test suite run flagged as an error by the test