[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 7 Mar 2014 19:59:41 -0800 (PST)
From: Anthony Tanoury <tanoury@...oo.com>
To: "john-users@...ts.openwall.com" <john-users@...ts.openwall.com>
Subject: Re: Help - mpi ocl restore session
Thanks for the response magnum!
On 14-03-07 02:31 AM, magnum wrote:
On 2014-03-07 07:26, Anthony Tanoury wrote:
>
>I use John the Ripper, version 1.8.0.2-bleeding-jumbo_mpi
>>[linux-x86-64-opencl]
>>
>Is this a very recent snapshot or an older one? Some timer oddities has
>changed (for the better, hopefully) very recently. Like yesterday...
>
my snapshot is about two weeks old.
I can run an mpirun opencl session just fine and all sessions
>>complete just fine. My only trouble is with session restore and only
>>if it involves remote hosts. I can resume a session if there is not a
>>remote host. However, if I terminate a session with a one or more
>>remote hosts using "killall mpirun", "kill HUP" or Ctrl-c and try to
>>restore the session, only one core or one GPU will resume.
>>
>When you say "one" I assume you mean only the root node resumes?
>
All 6 cores resume on the master node, but only one core on each of the
remote computers.
Does this happen even if the session had been running for 30 minutes or
>more? Did you set "Save = 60" in john.conf per the instructions? Before
>killing an MPI session, you should "kill -USR1" the mpirun that controls
>it. This should trigger a session save. Then wait at least 30 seconds
>before aborting them.
>
Yes, it happens even if the session has been running for more than 30
minutes.
I had "Save=600", but changed it to 60 in john.conf on all computers. I
did not notice any difference.
To abort, I use "pkill -USR1 mpirun" to trigger a session save, wait 30
seconds, then I do "killall mpirun". Is this the correct way to end a
session??
What happens with the other nodes? Do they silently just not resume or
>are there any errors or other clues?
>
The remote nodes resume but with only one core each.
When I start a new session I get this:
Device 2: Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
Device 1: Intel(R) Core(TM) i7 CPU X 980 @ 3.33GHz
Device 2: Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
Local worksize (LWS) 1, Global worksize (GWS) 8192
Local worksize (LWS) 1, Global worksize (GWS) 16384
Local worksize (LWS) 1, Global worksize (GWS) 1024
Loaded 4 password hashes with 4 different salts (wpapsk-opencl, WPA/WPA2
PSK [PBKDF2-SHA1 OpenCL 4x])
Node numbers 1-3 of 3 (MPI)
Note: minimum length forced to 8
Send SIGHUP to john process for status
However, when I try to resume a session I get this:
0 Session completed
0 Session completed
Device 1: Intel(R) Core(TM) i7 CPU X 980 @ 3.33GHz
Local worksize (LWS) 1, Global worksize (GWS) 16384
Loaded 4 password hashes with 4 different salts (wpapsk-opencl, WPA/WPA2
PSK [PBKDF2-SHA1 OpenCL 4x])
Node numbers 1-3 of 3 (MPI)
Note: minimum length forced to 8
Send SIGHUP to john process for status
Notice that only Device 1 of the master node is listed above, All six
cores on the master start, however, only core 2 on each of the remote
computers start.
If I do a "pkill -USR1 mpirun" after a session resume I will get:
--------------------------------------------------------------------------
mpirun noticed that process rank 1 with PID 7490 on node ub1 exited on
signal 10 (User defined signal 1).
--------------------------------------------------------------------------
and the session will abort and take me back to the prompt:
That message above dose not always indicate the same node but varies
between all nodes including the master.
Can you see any clues in the log files?
>
The john log files look good, no errors. Is there any other logs I
should check?
Also, the john.rec files on each computer are updated each time I do a
"pkill -USR1 mpirun" and look good.
If it's a very slow (unresponsive) format, try running with lower GWS
>(using eg. "mpirun -x GWS=2048" or whatever number is a lot lower than
>what is othereise used) when testing.
>
I lowered GWS down to 512 and no difference. Any more ideas?
magnum
>
>
>
Thanks again magnum!
On Friday, March 7, 2014 1:26 AM, Anthony Tanoury <tanoury@...oo.com> wrote:
Greetings JTR wizards-
I humbly bow before your knowledge again, in quest of jtr enlightenment....
I use John the Ripper, version 1.8.0.2-bleeding-jumbo_mpi [linux-x86-64-opencl]
I can run an mpirun opencl session just fine and all sessions complete just
fine. My only trouble is with session restore and only if it involves remote hosts. I can resume a session if there is not a remote host. However, if I terminate a session with a one or more remote hosts using
"killall mpirun", "kill HUP" or Ctrl-c and try to restore the session, only one core or one GPU will resume.
I use the following syntax to restore:
mpirun --host ub0 --host ub1 --host ub2 ./john --restore
I also have version 1.8.0.2-bleeding-jumbo_mpi [linux-x86-64-native] and I can
restore multi host sessions just fine usng:
mpirun -n 32 --hostfile /etc/nodes ./john --restore
Any idea why I'm having so much trouble restoring mpi opencl multi host sessions??
Thanks,
Tony
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
