[MPICH] setup and testing of smpd and mpiexec for MPICH2 under WinXP - Try specifying passphrase explicitly
Jayesh Krishna
jayesh at mcs.anl.gov
Tue Jan 15 11:44:23 CST 2008
Hi,
Can you try specify the passphrase explicitly and checking the status of
the process manager ? ("smpd -phrase behappy -status -d")
Regards,
Jayesh
_____
From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Kim Parnell
Sent: Monday, January 14, 2008 2:47 PM
To: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP - Is Reg enties ok ?
Some corrections on the Registry entries I sent earlier:
For:
values in HKEY_LOCAL_MACHINE\SOFTWARE\MPICH\SMPD
1st Entry: (Default) REG_SZ (value not set)
2nd Entry: binary REG_SZ
C:\MSC.Software\Marc\2007r1\marc2007r1\mpich2\bin\smpd.exe
3rd Entry: version REG_SZ 1.0.3
I do not have:
HKEY_CURRENT_USER\Software\MPICH\SMPD
or HKEY_CURRENT_USER\Software\MPICH\SMPD\CACHE
For
values in HKEY_CURRENT_USER\Software\MPICH
1st Entry: (Default) REG_SZ (value not set)
2nd Entry: smpdAccount REG_SZ kparnell
3rd Entry: smpdAccount1 REG_SZ NA\kparnell
4th Entry: smpdPassword REG_SZ (contains an encoded binary
string)
5th Entry: smpdPassword1 REG_SZ (contains an encoded binary
string)
_____
From: Kim Parnell
Sent: Monday, January 14, 2008 11:24 AM
To: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP - Is Reg enties ok ?
For:
values in HKEY_LOCAL_MACHINE\SOFTWARE\MPICH\SMPD
1st Entry: (Default) REG_SZ (value not set)
2nd Entry: binary REG_SZ
C:\MSC.Software\Marc\2007r1\marc2007r1\mpich2\bin\smpd.exe
3rd Entry: version REG_SZ 1.0.3
For:
values in HKEY_CURRENT_USER\Software\MPICH\SMPD\CACHE
1st Entry: (Default) REG_SZ (value not set)
2nd Entry: smpdAccount REG_SZ kparnell
3rd Entry: smpdAccount1 REG_SZ NA\kparnell
4th Entry: smpdPassword REG_SZ (contains an encoded binary
string)
5th Entry: smpdPassword1 REG_SZ (contains an encoded binary
string)
_____
From: Jayesh Krishna [mailto:jayesh at mcs.anl.gov]
Sent: Monday, January 14, 2008 10:43 AM
To: Kim Parnell
Cc: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP - Is Reg enties ok ?
Hi,
Can you see the registry entries for MPICH2 and check if they are *normal*
?
1. Run "regedit.exe"
2. Check the values in HKEY_LOCAL_MACHINE\SOFTWARE\MPICH\SMPD -- This
should contain binary (executable), phrase (passphrase) & version (version
number)
3. Check the values in HKEY_CURRENT_USER\Software\MPICH\SMPD\CACHE --- This
should contain the cached username(smpda)/password(smpdp) for smpd/mpiexec.
Regards,
Jayesh
_____
From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Kim Parnell
Sent: Friday, January 11, 2008 5:12 PM
To: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP
All of the results that I have reported thus far have been with MPICH2
uninstalled. Just using smpd.exe and wmpiexec.exe
It seems like the issue on this system is in the communication since
smpd -status
still does not return any info and has to be killed.
(I tested on a different but similar system with MPICH2 uninstalled and
returns a status)
On this system with the problem, the last lines returned by
smpd -status -d
....
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_do_console
(no message about the smpd process running, it just hangs here; my other
system returns a status here and a few additional lines)
_____
From: Jayesh Krishna [mailto:jayesh at mcs.anl.gov]
Sent: Friday, January 11, 2008 1:22 PM
To: Kim Parnell
Cc: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP
Hi,
Can you try a fresh installation (uninstall any existing version of MPICH2
on the machine and install MPICH2 again) of MPICH2
(mpich2-1.0.3-1-win32-ia32.msi) and let us know if it works ?
Regards,
Jayesh
_____
From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Kim Parnell
Sent: Friday, January 11, 2008 2:54 PM
To: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP
I can kill the smpd process. Then I can start it with:
smpd -start
and stop it with
smpd -stop
smpd -status
still does not return any info and has to be killed.
_____
From: Jayesh Krishna [mailto:jayesh at mcs.anl.gov]
Sent: Friday, January 11, 2008 12:45 PM
To: Kim Parnell
Cc: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP
Hi,
To restart smpd you should use "smpd -start" (not "smpd -install").
Regards,
Jayesh
_____
From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Kim Parnell
Sent: Friday, January 11, 2008 1:20 PM
To: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP
Let me provide the last items first:
Can you provide us the output of "smpd -status -d" ? Can you try "mpiexec
-n 1 dir" ?
mpiexec -n 1 dir
hangs at the command prompt and does not do a directory listing; if I
include -timeout 10 then it will timeout
smpd -status -d
[-1:5056]..\smpd_get_opt_int
[-1:5056]../smpd_get_opt_int
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt_int
[-1:5056]../smpd_get_opt_int
[-1:5056]..\smpd_get_opt_int
[-1:5056]../smpd_get_opt_int
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_hostname
[-1:5056]../smpd_get_hostname
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt_string
[-1:5056]../smpd_get_opt_string
[-1:5056]..\smpd_get_opt
[-1:5056]../smpd_get_opt
[-1:5056]..\smpd_do_console
^C to kill it (does not return to command prompt)
I just used "smpd -install" to restart the process after I had killed it
from the TaskManager.
I am not using the latest versions just to avoid possibly introducing
another variable with the MPICH2 applications that I ultimately want to run.
The version used is:
mpich2-1.0.3-1-win32-ia32.msi
There are not multiple versions and I running smpd and mpiexec from the
local directory.
Thanks!
_____
From: Jayesh Krishna [mailto:jayesh at mcs.anl.gov]
Sent: Friday, January 11, 2008 11:01 AM
To: Kim Parnell
Cc: mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP
Hi,
Are you using the latest version of MPICH2 (1.0.6p1)? I would recommend
that you try installing the latest version of MPICH2 on your machine. (Are
there multiple versions of smpd/mpiexec on your machine ? Check the PATH and
make sure that you are running the right smpd/mpiexec.)
Is there a reason you chose to run "smpd -install" (Instead of letting the
MPICH2 installer install SMPD as a service, which gets started when you
logon the next time)?
Can you provide us the output of "smpd -status -d" ? Can you try "mpiexec
-n 1 dir" ?
Regards,
Jayesh
_____
From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Kim Parnell
Sent: Friday, January 11, 2008 12:25 PM
To: mpich-discuss at mcs.anl.gov
Subject: [MPICH] setup and testing of smpd and mpiexec for MPICH2 under
WinXP
Windows XP SP2 machine with Dual Core CPU
I have an MPICH2 application that was working correctly and now does not get
submitted.
I am trying to go back and test the smpd and mpiexec installations and
hitting some problems that I do not understand.
smpd -install
starts an smpd process under user SYSTEM that seems to be running
smpd -status
starts another smpd process under my username but never returns a status;
basically just hangs at the command prompt
smpd -stop
Stopping MPICH2 Process Manager, Argonne National
Lab...................................
MPICH2 Process Manager, Argonne National Lab failed to stop.
I can kill the smpd process from the Windows Task Manager
I will put the Debug output from "smpd -d" below
mpiexec -validate -user USERNAME
hangs at the command prompt without any further output
mpiexec -register
prompts for username and password which I can register and is confirmed as
"Password encrypted into the Registry."
a simple test like:
mpiexec -timeout 20 -verbose -n 1 ping
will just terminate due to the timeout
I will put the verbose output below under the smpd debug
What am I missing? Thanks for any assistance.
Regards,
Kim
smpd -install -d
[-1:5076]..\smpd_get_opt_int
[-1:5076]../smpd_get_opt_int
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt_int
[-1:5076]../smpd_get_opt_int
[-1:5076]..\smpd_get_opt_int
[-1:5076]../smpd_get_opt_int
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
MPICH2 Process Manager, Argonne National Lab removed.
[-1:5076]..\smpd_get_opt_string
[-1:5076]../smpd_get_opt_string
[-1:5076]..\smpd_get_win_opt_string
[-1:5076]../smpd_get_win_opt_string
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
[-1:5076]..\smpd_get_opt_string
[-1:5076]../smpd_get_opt_string
[-1:5076]..\smpd_get_opt
[-1:5076]../smpd_get_opt
MPICH2 Process Manager, Argonne National Lab installed.
[-1:5076]..\smpd_set_smpd_data
[-1:5076]../smpd_set_smpd_data
C:\MSC.Software\Marc\2007r1\marc2007r1\mpich2\bin>mpiexec -timeout 20
-verbose -n 1 ping
..\smpd_get_full_path_name
...fixing up exe name: 'ping' -> '(null)'
../smpd_get_full_path_name
..handling executable:
ping.exe
..\smpd_get_next_host
...\smpd_get_next_hostname
.../smpd_get_next_hostname
...\smpd_get_host_id
.../smpd_get_host_id
../smpd_get_next_host
..\smpd_create_cliques
...\next_launch_node
.../next_launch_node
...\next_launch_node
.../next_launch_node
../smpd_create_cliques
..\smpd_fix_up_host_tree
../smpd_fix_up_host_tree
./mp_parse_command_args
.host tree:
. host: l30544.na.mscsoftware.com, parent: 0, id: 1
.launch nodes:
. iproc: 0, id: 1, exe: ping.exe
.\smpd_create_context
..\smpd_init_context
...\smpd_init_command
.../smpd_init_command
../smpd_init_context
./smpd_create_context
.\smpd_make_socket_loop
..\smpd_get_hostname
../smpd_get_hostname
./smpd_make_socket_loop
.\smpd_create_context
..\smpd_init_context
...\smpd_init_command
.../smpd_init_command
../smpd_init_context
./smpd_create_context
.\smpd_enter_at_state
..sock_waiting for the next event.
..SOCK_OP_CONNECT
..\smpd_handle_op_connect
...connect succeeded, posting read of the challenge string
../smpd_handle_op_connect
..sock_waiting for the next event.
..
mpiexec terminated job due to 20 second timeout.
mpiexec -timeout 100 -verbose -n 1 ping
..\smpd_get_full_path_name
...fixing up exe name: 'ping' -> '(null)'
../smpd_get_full_path_name
..handling executable:
ping.exe
..\smpd_get_next_host
...\smpd_get_next_hostname
.../smpd_get_next_hostname
...\smpd_get_host_id
.../smpd_get_host_id
../smpd_get_next_host
..\smpd_create_cliques
...\next_launch_node
.../next_launch_node
...\next_launch_node
.../next_launch_node
../smpd_create_cliques
..\smpd_fix_up_host_tree
../smpd_fix_up_host_tree
./mp_parse_command_args
.host tree:
. host: l30544.na.xxx.com, parent: 0, id: 1
.launch nodes:
. iproc: 0, id: 1, exe: ping.exe
.\smpd_create_context
..\smpd_init_context
...\smpd_init_command
.../smpd_init_command
../smpd_init_context
./smpd_create_context
.\smpd_make_socket_loop
..\smpd_get_hostname
../smpd_get_hostname
./smpd_make_socket_loop
.\smpd_create_context
..\smpd_init_context
...\smpd_init_command
.../smpd_init_command
../smpd_init_context
./smpd_create_context
.\smpd_enter_at_state
..sock_waiting for the next event.
..SOCK_OP_CONNECT
..\smpd_handle_op_connect
...connect succeeded, posting read of the challenge string
../smpd_handle_op_connect
..sock_waiting for the next event.
..
mpiexec terminated job due to 100 second timeout.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20080115/356af633/attachment.htm>
More information about the mpich-discuss
mailing list