Hi Mary and Dave.
Mary Haley ha scritto:
>
> Marco,
>
> Dave and I discussed this, and need more information.
>
>    - How many times is this procedure being called? Is it just
>      a few times, or 100s of times, or more than that?
More then 100s of time.
I will reduce the number of calls after this first period of monitoring.
>
>    - Are there any other error or warning messages appearing
>      before the fatal message?
>
Absolutely no
>    - Is the script you're running memory intensive in any way?
>      That is, are you doing something like opening and closing a lot
>      of files, creating large variables, running big do loops, or
>      doing a lot of variable reordering?
>
I read and delete big variable from a netcdf file, but not in the same 
time. The script reads variable and then deletes it.
The max RAM used is about 800 Mb.
"Do loops": yes there are some big (with a lot of instructions inside) 
do loops. They controls variable, time, and "regions", that i process.
>    - You said the script doesn't fail consistently, correct?
Yesterday (but just yesterday) it always has failed in the same point if 
i restart it. After the read and "reorganize" of a netcdf variable.
This is the part of the script:
------------------------------------------------
        _log("Netcdf: "+var+" is reading")
          
           nc_S_F          = 
new((/ixf,dimsizes(nc_zz_new),nc_ny_nx(0),nc_ny_nx(1)/),float)
       
             nc_S_F(:,1:(dimsizes(nc_zz_new)-2),:,:)          = 
(f->S(:,0:(dimsizes(nc_zz_new)-3),:,:))     
              nc_S_F(:,0,:,:)                                 = 
nc_S_F(:,1,:,:)
               nc_S_F(:,(dimsizes(nc_zz_new)-1),:,:)          = 
nc_S_F(:,(dimsizes(nc_zz_new)-2),:,:)
               nc_S_F!0                                       = "time"
                   nc_S_F!1                                       = "zpos"
                   nc_S_F!2                                       = "ypos"
                   nc_S_F!3                                       = "xpos"
                 nc_S_F_at_FillValue                               = 
nc_fillvalue
          
           nc_S_A          = 
new((/ixa,dimsizes(nc_zz_new),nc_ny_nx(0),nc_ny_nx(1)/),float)  
           
             nc_S_A(:,1:(dimsizes(nc_zz_new)-2),:,:)          = 
(a->S(:,0:(dimsizes(nc_zz_new)-3),:,:))
              nc_S_A(:,0,:,:)                                 = 
nc_S_A(:,1,:,:)
               nc_S_A(:,(dimsizes(nc_zz_new)-1),:,:)          = 
nc_S_A(:,(dimsizes(nc_zz_new)-2),:,:)
               nc_S_A!0                                       = "time"
                   nc_S_A!1                                       = "zpos"
                   nc_S_A!2                                       = "ypos"
                   nc_S_A!3                                       = "xpos"
                 nc_S_A_at_FillValue                               = 
nc_fillvalue
                    
        _log("Netcdf: "+var+" DONE")     
        end if
------------------------------------------------
>      Is it possible that the problem is with the load on your machine?
This was also my guess, but yesterday no other processes was running at 
the same time
>      That is, are there lots of other users running jobs on the
>      same system?
Just 3 users
>
> --Mary
>
>
> On Mon, 10 Nov 2008, Marco Pastore wrote:
>
>> Hi all.
>>
>> In my batch visualization script, that runs during the night to 
>> produce a lot of pictures, it happens sometimes this strange error:
>>
>> fatal:systemfunc: cannot create child process:[errno=12]
>>
>> during the call to these simple "log" procedure:
>>
>> ; add date and time string to console output
>> procedure _log(s:string)
>> begin
>> print(systemfunc("date +%F' '%T") + " " + s)
>> end
>>
>> The error is strange. In fact, if i restart the code, it can runs 
>> fine, or crash again. But i can not understand why.
>>
>> Do you have suggestions for me ?
>>
>> Thanks a lot, Marco
>>
>>
>> -- 
>> ###   ###   ### ### ###
>>
>> MARCO PASTORE
>>
>> Istituto Nazionale di Geofisica e Vulcanologia
>> Gruppo Nazionale di Oceanografia Operativa
>> Viale Aldo Moro 44
>> 40128, Bologna
>>
>> Tel:(+39)051-3782676
>> Cel:(+39)339 8273965
>>
>> Skype: mapas71
>>
>> ###   ###   ### ### ### _______________________________________________
>> ncl-talk mailing list
>> List instructions, subscriber options, unsubscribe:
>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>
>
>
-- 
###   ###   ### ### ###
MARCO PASTORE
Istituto Nazionale di Geofisica e Vulcanologia
Gruppo Nazionale di Oceanografia Operativa
Viale Aldo Moro 44
40128, Bologna
Tel:(+39)051-3782676
Cel:(+39)339 8273965
Skype: mapas71
    
###   ###   ### ### ### 
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Nov 11 2008 - 02:12:22 MST
This archive was generated by hypermail 2.2.0 : Thu Dec 11 2008 - 03:51:46 MST