[GE users] Reg: windows domain users access to grid

Harald Pollinger Harald.Pollinger at Sun.COM
Thu Nov 1 10:37:43 GMT 2007


    [ The following text is in the "ISO-8859-15" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

Hi Manju,

with the static_load.sh, do you still get the "load sensor died through 
signal = 11" messages in the execd messages file?

Regards,
Harald

manju a wrote:
> Hi Harald,
> 
> I changed the load sensors to static_load using this command qconf
> -mconf win_host_name n i restarted the execution host.. still not able
> to get that sge_execd process under process list!!!! and another thing
> i observed is qloadersensor.exe is running on the working
> machine(where the sge_execd is running)... but i couldn't find it in
> the not working machine... i reinstalled the SFU n applied the patch,
> yes off course  i disabled the DEP.... no luck yet.... this type of
> problem i m facing in nearly 5 to 6 machines... which is having two
> drives n user profile pointing in to D drive n checked the OS patch
> level it looks same in all the machines.........
> 
> thanks
> Manjunath
> 
> On Oct 31, 2007 10:40 PM, Harald Pollinger <Harald.Pollinger at sun.com> wrote:
>> Manju,
>>
>> signal 11 means "segmentation fault", i.e. the load sensor tries to
>> access some not allocated memory. As the load sensors works on all other
>> systems, I still assume that something is wrong with your Windows version...
>>
>> As a workaround, you could configure a different load sensor, at first
>> only a dummy load sensor. "qconf -sconf <windows_host_name>" shows you
>> what load sensor is currently configured, it should be
>> "$SGE_ROOT/util/resources/loadsensors/interix-loadsensor.sh".
>>
>> Just configure a different load sensor script, e.g. one that only
>> returns static data like the static_load.sh I've attached to this E-Mail.
>>
>> If the execd works with this load sensor, we should dig deeper into the
>> load sensors problem.
>>
>> Regards,
>> Harald
>>
>>
>>
>>
>> manju a wrote:
>>> Hi Harald,
>>>
>>> please find the logs from the execution host (this logs from windows
>>> execution host)
>>>
>>> 10/31/2007 06:23:29|execd|execd_host|I|starting up SGE 6.1u2 (win32-x86)
>>> 10/31/2007 06:23:31|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:23:33|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:23:35|execd|execd_host|I|controlled shutdown 6.1u2
>>> 10/31/2007 06:24:04|execd|execd_host|I|starting up SGE 6.1u2 (win32-x86)
>>> 10/31/2007 06:24:05|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:24:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:24:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:25:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:26:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:26:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:27:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:28:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:28:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:29:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:30:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:30:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:31:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:32:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:32:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:33:27|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:34:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:34:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:35:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:36:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:36:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:37:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:38:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:38:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:39:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:40:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:40:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:41:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:42:06|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:42:46|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:43:26|execd|execd_host|E|load sensor died through signal = 11
>>> 10/31/2007 06:43:29|execd|execd_host|I|controlled shutdown 6.1u2
>>>
>>> if i tried to start from $SGE_ROOT/bin/win32-x86/sge_execd after i m
>>> in to dl 5 mode.... it will keep on struck at starting sge_execd , at
>>> that time i can see the sge_execd process under process list !!!! but
>>> it will not start.....
>>>
>>> please let me know if you got any thing atleast with this logs....
>>>
>>> thanks
>>> Manjunath A.
>>>
>>>
>>>
>>> On Oct 31, 2007 7:00 PM, manju a <manju.kudu at gmail.com> wrote:
>>>> yes i done that one also!!!! but no luck..... if i start
>>>> /etc/init.d/sgeexecd start process it will come n disappear within a
>>>> sec from the process list....
>>>>
>>>>
>>>> On Oct 31, 2007 6:52 PM, Harald Pollinger <Harald.Pollinger at sun.com> wrote:
>>>>> manju a wrote:
>>>>>> Hi Harald,
>>>>>>
>>>>>> i already applied this hot fixes but not as suggested in the page(i
>>>>>> mean order while applying the hot fixes) does it makes any difference
>>>>>> ???
>>>>> The page says: Yes, it makes a difference, apply them only in the
>>>>> suggested order to make sure all of them work.
>>>>>
>>>>> Regards,
>>>>> Harald
>>>>>
>>>>>
>>>>>> thanks
>>>>>> Manju
>>>>>>
>>>>>> On Oct 31, 2007 4:56 PM, Harald Pollinger <Harald.Pollinger at sun.com> wrote:
>>>>>>> manju a wrote:
>>>>>>>> Hi Harald,
>>>>>>>>
>>>>>>>>  i can see the below messages from master at  this location
>>>>>>>> $SGE_ROOT/farm/spool/qmaster/messages
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> 10/30/2007 23:00:07|qmaster|testqmaster|E|commlib error: got read
>>>>>>>> error (closing "execd_host.abc.com/qstat/85")
>>>>>>>> 10/30/2007 23:28:44|qmaster|testqmaster|E|commlib error: got read
>>>>>>>> error (closing "execd_host.abc.com/qconf/121")
>>>>>>>> 10/30/2007 23:32:03|qmaster|testqmaster|E|commlib error: got read
>>>>>>>> error (closing "execd_host.abc.com/qconf/125")
>>>>>>>> 10/30/2007 23:32:44|qmaster|testqmaster|E|commlib error: got read
>>>>>>>> error (closing "execd_host.abc.com/execd/1")
>>>>>>>> 10/30/2007 23:33:28|qmaster|testqmaster|E|commlib error: got read
>>>>>>>> error (closing "execd_host.abc.com/execd/1")
>>>>>>>>
>>>>>>>> what this commlib error means ???
>>>>>>> It means that the qmaster is just before reading or already is reading
>>>>>>> some data from the execd, but then the execd suddenly 'vanishes'.
>>>>>>>
>>>>>>>
>>>>>>>> testqmaster ----> qmaster server name
>>>>>>>>
>>>>>>>> execd_host-----> execution host name (sge_execd process not running
>>>>>>>> machine name )
>>>>>>>>
>>>>>>>> pleas let me know if you got any thing from this logs....i m still
>>>>>>>> trying to make run this sge_execd process but no luck.. please help me
>>>>>>>> out this.......
>>>>>>> I guess there is simply something wrong with your Windows/SFU host. You
>>>>>>> could try to apply all Hotfixes suggested on this page:
>>>>>>> http://www.duh.org/interix/hotfixes.php
>>>>>>>
>>>>>>> This is just a collection of all official Microsoft Hotfixes for SFU.
>>>>>>> It's very difficult to find the Hotfixes and their dependencies on the
>>>>>>> Microsoft pages, so just apply them in the order suggested here for your
>>>>>>> Windows version.
>>>>>>>
>>>>>>> Regards,
>>>>>>> Harald
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> thanks
>>>>>>>> Manjunath A,
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Oct 30, 2007 4:45 PM, Harald Pollinger <Harald.Pollinger at sun.com> wrote:
>>>>>>>>> manju a wrote:
>>>>>>>>>> Hi Harald,
>>>>>>>>>>
>>>>>>>>>> i like to give some more information
>>>>>>>>>>
>>>>>>>>>> whatever working machines all the windows profiles pointing in to
>>>>>>>>>> "C:\Document and Settings"  do you think this will matters ????
>>>>>>>>>>
>>>>>>>>>> and non working machines (sge_execd not running) Windows profiles
>>>>>>>>>> pointing to "D:\Users"
>>>>>>>>>>
>>>>>>>>>> this is the only difference i found between working and non working machines...
>>>>>>>>>>
>>>>>>>>>> no error found while i m installing machine as execution host, i
>>>>>>>>>> carried out the installation process same as the working machines....
>>>>>>>>> This might matter for job execution, but it doesn't matter for the Execd
>>>>>>>>> startup.
>>>>>>>>>
>>>>>>>>> I found this:
>>>>>>>>> The "TERMINATE dispatcher because j == 1" line gets printed when the
>>>>>>>>> Execd either receives a SIGTERM or a SIGINT signal or it receives a
>>>>>>>>> shutdown request from the QMaster.
>>>>>>>>>
>>>>>>>>> As I can't see any request from the QMaster in the trace file sniplet
>>>>>>>>> you sent me, I assume the Execd received a SIGTERM or a SIGINT signal.
>>>>>>>>> I can't tell you where this signal came from.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Regards,
>>>>>>>>> Harald


-- 
Sun Microsystems GmbH         Harald Pollinger
Dr.-Leo-Ritter-Str. 7         N1 Grid Engine Engineering
D-93049 Regensburg            Phone: +49 (0)941 3075-209  (x60209)
Germany                       Fax: +49 (0)941 3075-222  (x60222)
http://www.sun.com/gridware
mailto:harald.pollinger at sun.com
Sitz der Gesellschaft: Sun Microsystems GmbH, Sonnenallee 1,
D-85551 Kirchheim-Heimstetten
Amtsgericht Muenchen: HRB 161028
Geschaeftsfuehrer: Wolfgang Engels, Dr. Roland Boemer
Vorsitzender des Aufsichtsrates: Martin Haering

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list