Opened 6 months ago

Closed 4 weeks ago

#2939 closed help (answered)

Calendar type gregorian not recognised

Reported by: yaweiqu Owned by: um_support
Component: UKESM Keywords:
Cc: Platform: Monsoon2
UM Version: 11.2

Description

Hi,

I am using UM11.2 UKESM to do some nudging simulations. I'm trying to convert a 360day suite (u-bj793) to Gregorian and active nudging (u-bj856). The suite u-bj856 crashed with error as follows:

[WARN] file:STASHC: skip missing optional source: namelist:exclude_package(:)
[WARN] file:ATMOSCNTL: skip missing optional source: namelist:jules_urban2t_param
[WARN] file:RECONA: skip missing optional source: namelist:trans(:)
[WARN] file:IDEALISE: skip missing optional source: namelist:idealised
[WARN] file:RECONA: skip missing optional source: namelist:ideal_free_tracer(:)
[WARN] file:IOSCNTL: skip missing optional source: namelist:lustre_control
[WARN] file:IOSCNTL: skip missing optional source: namelist:lustre_control_custom_files
[WARN] file:RECONA: skip missing optional source: namelist:recon_idealised
[WARN] file:SHARED: skip missing optional source: namelist:jules_urban_switches
[FAIL] setup_runtime: Calendar type gregorian not recognised
[FAIL] run_model # return-code=103
2019-06-19T05:06:49Z CRITICAL - failed/EXIT

The changes I've made including https://code.metoffice.gov.uk/trac/roses-u/changeset/120422 and https://code.metoffice.gov.uk/trac/roses-u/changeset/120366.

Could you please help me to solve this issue? Thank you in advance.

Best,
Yawei

Change History (7)

comment:1 Changed 6 months ago by yaweiqu

Sorry. The old 360day suite is u-bi793.

comment:2 Changed 6 months ago by grenville

Yawei

The suite sets CALENDAR="gregorian", but the driver is checking for

calendar = runtime_envarCALENDAR?
elif calendar == '365day':

so it fails - this may be fixed somewhere. We'll need to check.

Grenville

comment:3 Changed 6 months ago by grenville

Yawei

You could change the code in

/home/d03/yaweiqu/cylc-run/u-bj856/share/fcm_make_drivers/build/bin/common.py

to refer to gregorian rather than 356day.

Are you in contact with the UKESM team - changing calendars may not be a simple operation; they should be able to advise.

Grenville

comment:4 Changed 6 months ago by yaweiqu

Hi Grenville,

Thanks for your reply.

Sorry I haven't contacted the UKESM team regarding this calendar problem. I'm not sure who would know the solution to this so I just created the help ticket.

I also found a similar 365day setting in /home/d03/yaweiqu/roses/u-bj856/tests-runtime.rc.

DAYS_ELAPSED_AT_CYCLE= $(rose date --calendar=365day ${CYLC_SUITE_INITIAL_CYCLE_POINT} ${CYLC_TASK_CYCLE_POINT} -f d)

I changed the 365day to gregorian in both common.py and tests-runtime.rc. Then I reloaded the suite u-bj856. This 3-month test succeeded.

However, if I run the suite with rose suite-run —new instead of —reload, the suite fails again. I think this might because /home/d03/yaweiqu/cylc-run/u-bj856/share/fcm_make_drivers/build/bin/common.py is a file copied from the source code for every cycle run. It will remain '365day' for each new run if I don't change the source code.

Do you know how to change the source code, or whom (UKESM team) should I contact?

Many thanks,
Yawei

comment:5 Changed 5 months ago by grenville

Hi Yawei

You can change the source code - we run a training course which covers this and much more. The training material is available, see http://cms.ncas.ac.uk/wiki/UmTraining/April2019 for presentations and practicals. We will be running a course later in the year.

I shall update the UKESM team re 365day and gregorian.

rose suite-run —new does rebuild everything - but rarely needs to be used.

Grenville

comment:6 Changed 5 months ago by yaweiqu

Hi Grenville,

Thanks for the info.

Since u-bj856 (3-mon test) succeeded, I copied it to u-bk172 and changed the runtime to 5 years. But u-bk172 failed at the begining of the second year (2015) with error:

????????????????????????????????????????????????????????????????????????????????
???!!!???!!!???!!!???!!!???!!!       ERROR        ???!!!???!!!???!!!???!!!???!!!
?  Error code: 1983
?  Error from routine: GET_EMFILE_REC
?  Error message: Invalid time record 2, End-of-file reached ?
?  Error from processor: 540
?  Error number: 322
????????????????????????????????????????????????????????????????????????????????

[540] exceptions: An non-exception application exit occured.
[540] exceptions: whilst in a serial region
[540] exceptions: Task had pid=27970 on host nid06692
[540] exceptions: Program is "./atmos.exe"
Warning in umPrintMgr: umPrintExceptionHandler : Handler Invoked
Rank 540 [Sun Jun 30 16:50:52 2019] [c6-2c2s9n0] application called MPI_Abort(MPI_COMM_WORLD, 9) - process 540
Rank 558 [Sun Jun 30 16:50:52 2019] [c6-2c2s9n1] application called MPI_Abort(MPI_COMM_WORLD, 9) - process 558
atpAppSigHandler: Back-end never delivered its pid. Re-raising signal.
_pmiu_daemon(SIGCHLD): [NID 06551] [c6-2c0s5n3] [Sun Jun 30 16:54:58 2019] PE RANK 6 exit signal Aborted
[NID 06551] 2019-06-30 16:54:58 Apid 69290765: initiated application termination
[FAIL] run_model # return-code=137
2019-06-30T16:55:15Z CRITICAL - failed/EXIT

The suite u-bk172 (or u-bj856) is a nudging version of u-bi793. u-bi793 had run 5-years and ended without error, so I think the error might occur because of the changes I've made:
https://code.metoffice.gov.uk/trac/roses-u/changeset/120366
https://code.metoffice.gov.uk/trac/roses-u/changeset/120422
https://code.metoffice.gov.uk/trac/roses-u/changeset/121871.

I've seen the GET_EMFILE_REC error in #2679 and #2787 - it's because some of my emission files have the wrong time. However, there's no difference in emission files between u-bi793 and u-bk172, except for the files compatible with Gregorian calendar, such as:

easyaerosol_dir='/projects/ukca-admin/inputs/ancil/n96e/cmip6/VolcanicAod/tseries_1850-2014/365d'
...
ukca_em_files='/projects/ukca-admin/inputs/ancil/n96e/cmip6/AeroChemEms/clim_2000/v1/aerocom/ukca_emiss_SO2_nat_grg.nc',
...
             ='/projects/ukca-admin/inputs/ancil/n96e/cmip6/AeroChemEms/tseries_1850-2014/CH4_biomass_low_1849_2015_time_series_grg.nc',

The error might be related to the ancil files above, because some of them are 1850-2014 while my simulation is 2014-2019. I don't know if there are other files with proper time series that can be used for my nudging/Gregorian suite. Do you have any idea about this?

Cheers,
Yawei

Last edited 5 months ago by yaweiqu (previous) (diff)

comment:7 Changed 4 weeks ago by ros

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