Opened 3 years ago

Closed 3 years ago

#1977 closed help (fixed)

UKCA run error

Reported by: emxin Owned by: um_support
Component: UKCA Keywords:
Cc: Platform: ARCHER
UM Version: 7.3

Description

hi there,

I met a problem when I tried to output model result in a 3hrly frequency. UKCA crashed with cores created. The error messages I got are related to 'section=39, item =1'(see below). I could not find 'section=39', can you help? Thanks.

OPEN: File /work/n02/n02/emxin/um/xkqjs/xkqjs.astart to be Opened on Unit 21 Exists

gc_abort (Processor 18 ): PRELIM:TOTIMP:Error in time period conversion model= 1 section= 39 item= 1


and things like:

Process 1 has exited.

*
UM ERROR (Model aborting) :
Routine generating error: UM_SHELL
Error code: 101
Error message:

PRELIM:TOTIMP:Error in time period conversion model= 1 section= 39 item= 1

cheers
Xin

Change History (5)

comment:1 Changed 3 years ago by luke

Hi Xin,

Section 39 is nudging, so I'm assuming that you're using a nudged model.

How have you requested 3 hourly output, and which variables have you requested?

Currently only 4 diagnostics are requested from xkqjs:

34   1 O3 MASS MIXING RATIO  AFTER TSTEP   T6H      DALLTH   UPC      Y +  Y  
34 153 BrO MASS MIXING RATIO AFTER TSTEP   T6H      DALLTH   UPC      Y +  Y  
34 352 COLUMN BRO:TROP ONLY                T6H      DALLTH   UPJ      Y +  Y  
34 354 COLUMN BRO: WHOLE ATMOS             T6H      DALLTH   UPJ      Y +  Y  

all 6-hourly it seems. Have you turned diagnostics off after you raised this ticket?

Many thanks,
Luke

comment:2 Changed 3 years ago by emxin

hi Luke,
Yes, I have changed the diagnostics in xkqjs after I raised the ticket. Since I raised the ticket, I have done two test runs: 1) kept the 3hrly output request (in jobID xkqju); 2) changed the 3hrly output request back to the 6hrly one (in jobID xkqjt). However, both failed with core files formed (two cores in xkqjt, and one core in xkqju).

In both runs, the error messages for section 39 has gone, which is good. The error in both runs seems related to item 34352 (column BrO). in the STASH panel, the options for /time/domain/usage/ are /T6H/DALLTH/UPJ/ in xkqjt and /T3H/DALLTH/UPJ/ in xkqju. Maybe I should use 'DIAG' for the domain? Thank for help in advance?

For your information, the error messages are:

 Leaving FLIGHT_OUT
  -@▒▒^?^@^@P^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@~]▒▒▒▒▒▒▒^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@ 64
 *********************************************************************************
 UM ERROR (Model aborting) :
 Routine generating error: ASAD_FLUX_PUT_STASH
 Error code:  34352
 Error message:
 -@▒▒^?^@^@P^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^

Xin

Last edited 3 years ago by luke (previous) (diff)

comment:3 Changed 3 years ago by luke

Hi Xin,

This looks a bit like #1182 . Please try the fix here:

http://www.ukca.ac.uk/wiki/index.php/Bugfixes#Cray_compiler_on_ARCHER

and see if that works.

What is the structure of the BrO column array? If it's a 3D field then DALLTH should be OK, although you might just want the lowest level scientifically. DIAG should be used for 2D arrays. If you want the lowest level of the 3D array you'll need to use a different domain, where you only output the lowest theta level. Looking in this job you'll need to make this domain profile yourself as one isn't present.

Thanks,
Luke

comment:4 Changed 3 years ago by emxin

Hi Luke,

I followed the instruction above and checked the code in asad_chem_flux_diags.F90 and found the few lines that may cause the problem have already been commented out. since I noticed that there is one more line was commented out in the code( icode=(1000*section) + item), thus, I corrected it and submit the job again to see what I can get.

In terms of the column BrO, it is a 3D field (the BrO is actually for its own gridbox column, instead of the total column BrO above). thus 3D output is needed. I know what you mean regarding to the 'diag' option.

Xin

comment:5 Changed 3 years ago by luke

  • Resolution set to fixed
  • Status changed from new to closed

Closing this ticket as Xin has reported that the problem seems to have been fixed with the above changes.

Note: See TracTickets for help on using tickets.