F compset requiring docn pointer files for second advance

[thoar]

I don't get it ... I have a multi-instance F compset using a single data ocean for all instances.I run for one day, everything seems perfect.I change CONTINUE_RUN = TRUE and the resubmission fails with the following error: (docn_comp_init) ocn_in = docn_ocn_in
(docn_comp_init) decomp = 1d
(docn_comp_init) restfilm = undefined
(docn_comp_init) restfils = undefined
(shr_strdata_readnml) reading input namelist file: docn_ocn_in
(shr_stream_init) Reading file docn.streams.txt.prescribed
(shr_file_freeUnit) unit 94 was not in use
(docn_comp_init) ocn mode = SSTDATA
(shr_strdata_init) calling shr_dmodel_mapSet for fill
(shr_strdata_init) calling shr_dmodel_mapSet for remap
(shr_strdata_print) ----------------------------------------------------------
(shr_strdata_print) name = SDOCN data
(shr_strdata_print) dataMode = SSTDATA
(shr_strdata_print) domainFile = /glade/p/cesmdata/cseg/inputdata/share/domains/domain.ocn.gx1v6.090206.nc
(shr_strdata_print) nxg = 320
(shr_strdata_print) nyg = 384
(shr_strdata_print) calendar = GREGORIAN
(shr_strdata_print) io_type = 5
(shr_strdata_print) eccen = 1.000000E+36
(shr_strdata_print) mvelpp = 1.000000E+36
(shr_strdata_print) lambm0 = 1.000000E+36
(shr_strdata_print) obliqr = 1.000000E+36
(shr_strdata_print) nstreams = 1
(shr_strdata_print) pio_iotype = 5
(shr_strdata_print) streams ( 1) = docn.streams.txt.prescribed 1850 1850 2012
(shr_strdata_print) taxMode ( 1) = cycle
(shr_strdata_print) dtlimit ( 1) = 1.500000E+00
(shr_strdata_print) strnxg ( 1) = 288
(shr_strdata_print) strnyg ( 1) = 192
(shr_strdata_print) dofill ( 1) = T
(shr_strdata_print) fillalgo( 1) = nn
(shr_strdata_print) fillmask( 1) = nomask
(shr_strdata_print) fillread( 1) = NOT_SET
(shr_strdata_print) fillwrit( 1) = NOT_SET
(shr_strdata_print) domaps ( 1) = T
(shr_strdata_print) mapalgo ( 1) = bilinear
(shr_strdata_print) mapmask ( 1) = dstmask
(shr_strdata_print) mapread ( 1) = NOT_SET
(shr_strdata_print) mapwrit ( 1) = NOT_SET
(shr_strdata_print) tintalgo( 1) = linear
(shr_strdata_print)
(shr_strdata_print) nvectors = 0
(shr_strdata_print) ----------------------------------------------------------
(docn_comp_init) initialize gsmaps
(docn_comp_init) copy domains
(docn_comp_init) allocate AVs
(docn_comp_init) restart filenames from rpointer
(docn_comp_init) ERROR: rpointer file does not exist
(shr_sys_abort) ERROR: (docn_comp_init) ERROR: rpointer file missing
(shr_sys_abort) WARNING: calling shr_mpi_abort() and stopping I know how to make a pointer file, but what the heck would I put in it for a DOCN?And why does the DOCN want a pointer file anyway?My CASEROOT directory is /glade/p/work/thoar/cases/cam_testMy RUN directory is /glade/scratch/thoar/cam_test/runAny help would be good - Thanks -- Tim

 

[thoar]

I forgot to mention I am running cesm1_1_1 with no docn sourcemods.

 

[thoar]

Following a hunch from Andy Mai and Nancy Norton (all cases were run with ocn_ncpl = 4/day) :an initial 48 hour advance DOES produce an rpointer.ocn file as well as cam_test.docn.rs1.2004-01-10-43200.bin file.an initial 24 hour advance DOES produce an rpointer.ocn file and a .rs1. file.an initial 12 hour advance DOES produce an rpointer.ocn file and a .rs1. file.an initial 6 hour advance fails to produce a pointer file or a .rs1. file. The contents of the pointer file are:cam_test.docn.r.2004-01-10-43200.nc                                                                                                                                                                                                                           cam_test.docn.rs1.2004-01-10-43200.bin                                                                                                          

 

[thoar]

Red Herring. Even with OCN_NCPL = 48/day, a 6hour advance fails to produce a pointer file or a .rs1. file.Feels like a bug to me.

 

[thoar]

Resolution: the F compset is 'tightly coupled' with the (data) ocean.If I accept the default values for env_run.xml:xxx_NCPL - and have STOP_N = 6 hours,with CONTINUE_RUN=FALSE ... an rpointer.ocn file IS created.I had specified a different OCN_NCPL frequency to match a separate compset I wanted to compare.To be specific ... I had built the case with OCN_NCPL = 4 (i.e. not the default value) ... and the runs fail.Taking that failed run and changing the OCN_NCPL back to the default value seemed to fail (the comment from 2 weeks ago).Building the case with the default OCN_NCPL ... results in a case that works.