Debug print statement causing large e3sm log files in elm-betr

[ndkeen]

Test ERS.f19_g16.I1850ELM.gcp12_gnu.elm-betr now writes 2G e3sm log file.
Over 37M lines that look like:

27:  crop          25          17           1           0
27:  crop          25          17           1           0
27:  crop          25          17           1           0
27:  crop          25          17           1           0

Looks to be here src/external_models/sbetr/src/driver/shared/BeTRSimulation.F90

    if(present(pft))then
      this%betr_pft(c)%column(:)=1
      this%betr_pft(c)%npfts = 0
      pp = 0
      do pi = 1, betr_maxpatch_pft
        if (pi <= col%npfts(c)) then
          p = col%pfti(c) + pi - 1
          if (pft%active(p) .and. (pft%itype(p) /= noveg)) then
            pp = pp + 1
            this%betr_pft(c)%wtcol(pp) = pft%wtcol(p)
            this%betr_pft(c)%itype(pp) = pft%itype(p)
            this%betr_pft(c)%crop(pp) = crop(pi)         !the crop looks weird here, jyt                                                                                                                                                               
          endif
        endif
      enddo
      this%betr_pft(c)%npfts = pp
      this%betr_pft(c)%crop(0:betr_maxpft)=crop(0:betr_maxpft)
      print*,'crop',size(crop),betr_maxpatch_pft,col%npfts(c), this%betr_pft(c)%npfts ! <----
    endif

[ndkeen]

I can't find the PR that introduced this, perhaps because it's in external repo. I've been seeing issues with e3sm_integration for at least 2 days, but perhaps a few more. The disk fills up 100% and strange things happen, with not all tests reporting.

perlmutter-login35% pwd
/global/cfs/cdirs/e3sm/ndk/repos/me56-nov6/components/elm/src/external_models/sbetr
perlmutter-login35% git blame src/driver/shared/BeTRSimulation.F90 | grep print | grep crop
96e3d3186 src/driver/shared/BeTRSimulation.F90 (Jinyun Tang     2023-08-18 15:39:29 -0700 2271)       print*,'crop',size(crop),betr_maxpatch_pft,col%npfts(c), this%betr_pft(c)%npfts

@jinyuntang

[ndkeen]

OK grepping thru older checkouts I have, the print in question was there as early as April. So while we still may want to remove this (or put under DEBUG?), it can't be responsible for the sudden increase in disk space by integration tests on gcp12. I did verify that commenting out that line does allow the e3sm_integration suite to complete (though I think I also removed the netcdf files from the extra_coverage tests -- which isn't much).

Unless... something changed that caused a new code path to be executed.
With a July2 checkout, I tried the test and see the massive number of crop output lines, so it's been there a while.

[mahf708]

#6214

[jinyuntang]

@ndkeen @rljacob I will look into this today.