I am trying to use the MPI capabilities of john 1.7.9 and am testing with a simple 4-char password on a single 8-core server machine. I noticed that after success, the process that cracked the password exits, while the rest remain running. Is this a known issue, or a new bug? Also, is there a workaround (apart from monitoring the time of each process and looking for differences, which is not really conclusive).
Details:
mpiexec --version
mpiexec (OpenRTE) 1.4.3
...
mpiexec -n 8 -host localhost ./run/john temp --format=raw-md5
...
mpiexec -n 8 -host localhost killall -s HUP john && mpiexec -n 8 -host localhost ./john --status
0: guesses: 0 time: 0:00:00:07 0.00% (3) c/s: 4424K
3: guesses: 0 time: 0:00:01:35 0.00% (3) c/s: 7985K
7: guesses: 0 time: 0:00:01:34 0.00% (3) c/s: 7190K
4: guesses: 0 time: 0:00:01:35 0.00% (3) c/s: 6643K
6: guesses: 0 time: 0:00:01:35 0.00% (3) c/s: 7319K
1: guesses: 0 time: 0:00:01:34 0.00% (3) c/s: 9367K
2: guesses: 0 time: 0:00:01:34 0.00% (3) c/s: 6774K
5: guesses: 0 time: 0:00:01:35 0.00% (3) c/s: 6740K
SUM: guesses: 0 time: 0:00:01:35 0.00% (3) c/s: 59635K avg 7454K
Notice process 0's running time and how it still reports 0 guesses. Running john --status shows the following:
./run/john --show temp --format=raw-md5
?:asdq
1 password hash cracked, 0 left
