Opened 9 years ago

Closed 9 years ago

#807 closed help (fixed)

REPLANCA: PP HEADERS ON ANCILLARY FILE DO NOT MATCH

Reported by: jonny Owned by: annette
Component: UM Model Keywords: HadGEM1.2 pp headers
Cc: Platform:
UM Version: 6.1

Description

Hi,
My HadGEM1.2 job xgyuf is failing at the start of the run. I am getting the following error message:
*

UM ERROR (Model aborting) :
Routine generating error: UP_ANCIL
Error code: 323
Error message:

REPLANCA: PP HEADERS ON ANCILLARY FILE DO NOT MATCH

*

I have looked at all the ancillary files and the dump files and can't see anything wrong. The information in the .leave file (xgyuf000.xgyuf.d12069.t093543.leave) does not give much of a clue as to which file is causing a problem. Any ideas about how I can diagnose the problem and flag up what the offending ancillary file is?

Thanks
Jonny

Change History (3)

comment:1 Changed 9 years ago by annette

  • Owner changed from um_support to annette
  • Status changed from new to accepted

Hi Jonny,

The code is crashing out in subroutine REPLANCA. I searched for the error message "PP HEADERS ON ANCILLARY FILE DO NOT MATCH" in the 6.1 code browser (as I don't have access to your source code):

http://cms.ncas.ac.uk/code_browsers/UM6.1/UMbrowser/index.html

And I found where the error message is written:

        IF(LMISMATCH) THEN                                                 RPANCA1A.729 
          ICODE=200+FIELD                                                  RPANCA1A.730 
         CMESSAGE='REPLANCA: PP HEADERS ON ANCILLARY FILE DO NOT MATCH'    RPANCA1A.731 
         RETURN                                                            RPANCA1A.732 
        END IF                                                             RPANCA1A.733 

From this we can decode that the field it's having problems with is 123 (i.e. 323-200).
Looking a bit further at the code just before this, the problem is a mismatch between the expected STASH code and STASH code in the ancillary file.

Skimming through your .leave file I found that it's using the following reference file for ancillary fields:

/work/n02/n02/hum/vn6.1/ctldata/ANCILmaster/ANCILfields_A

Looking at this file we can see that ancil field 123 corresponds to the following:

1|  123 |    1 |  132 |DMS concentration in seawater       |
2|   29 |

Next if we look at the DMS conecntration of seawater file from your job we can see that the STASH code in the file is 301, which does not match the expected STASH code 132. So this is what is causing the problem.

By the way, you can generate extra output for debugging purposes in the UMUI panel Sub-Model Independent → Output Options which can sometimes help pinpoint this type of problem. This can produce a lot of extra output so you will probably want to switch back for longer runs.

I hope this helps

Annette

comment:2 Changed 9 years ago by jonny

Hi,

Thanks for explaining the root of the problem to me. I was slightly confused as this variable DMS concentration is defined twice. Once with stash code 132 in the ANCILfields_A file as you describe, then once in the

~umui/hadgem1/vn6.1/preSTASHmaster/pre_DMS_dummy_601

file, I guess this is the legacy of development with this variable model. But this runs fine now.

Thanks
Jonny

comment:3 Changed 9 years ago by annette

  • Resolution set to fixed
  • Status changed from accepted to closed
Note: See TracTickets for help on using tickets.