mis-match between rpointer filename and what the model is expecting

[islasimpson]

What happened?

I'm using cam6_4_097 and I'm doing a branch run and using the following to grab the restart from a previous case but it's crashing because it looks for rpointer.cice instead of rpointer.cice.2004-01-01-00000.

./xmlchange RUN_TYPE="branch"
./xmlchange RUN_REFCASE="f.e30_cam6_4_097.FHISTC_MTso.ne30pg3_ne30pg3_mg17.NUDGED_UVT_dp0.15_scaling0.32.002"
./xmlchange RUN_REFDIR="/glade/derecho/scratch/islas/archive/f.e30_cam6_4_097.FHISTC_MTso.ne30pg3_ne30pg3_mg17.NUDGED_UVT_dp0.15_scaling0.32.002/rest/2004-01-01-00000/"
./xmlchange RUN_REFDATE="2004-01-01"
./xmlchange GET_REFCASE="TRUE"

I have fixed the problem with help from @cecilehannay by setting the pointer_file in user_nl_cice. But it seems like this could frequently trip people up. Is this expected behavior or a bug?

What are the steps to reproduce the bug?

In cam6_4_097

./create_newcase --compset FHISTC_MTso --res ne30pg3_ne30pg3_mg17 --case $CASEDIR --run-unsupported --project $PROJECT
cd $CASEDIR
./xmlchange RUN_STARTDATE="2004-01-01"
./xmlchange STOP_OPTION=nyears
./xmlchange STOP_N=2
./xmlchange RESUBMIT=0
./xmlchange RUN_TYPE="branch"
./xmlchange RUN_REFCASE="f.e30_cam6_4_097.FHISTC_MTso.ne30pg3_ne30pg3_mg17.NUDGED_UVT_dp0.15_scaling0.32.002"
./xmlchange RUN_REFDIR="/glade/derecho/scratch/islas/archive/f.e30_cam6_4_097.FHISTC_MTso.ne30pg3_ne30pg3_mg17.NUDGED_UVT_dp0.15_scaling0.32.002/rest/2004-01-01-00000/"
./xmlchange RUN_REFDATE="2004-01-01"
./xmlchange GET_REFCASE="TRUE"

What CAM tag were you using?

cam6_4_097

What machine were you running CAM on?

CISL machine (e.g. cheyenne)

What compiler were you using?

Intel

Path to a case directory, if applicable

No response

Will you be addressing this bug yourself?

No

Extra info

No response

[cacraigucar]

@islasimpson - Please try using a branch we are getting ready to commit for CAM and see if this is still a problem:

git clone https://github.com/cacraigcuar/CAM cam_alpha07d
cd cam_alpha07d
git checkout cam_match_alpha07d_externals

Please post your results back here in this issue

[islasimpson]

I'm trying with this alpha tag but I can't compile. I get an error "interpolate_output does not work with COSP". I have interpolate_output = .true. in user_nl_cam, but I'm not setting anything related to COSP. I'm running the same script as I was using with cam6_4_097 and only changed the tag. Do I now have to actively turn COSP off?

[gold2718]

I'm trying with this alpha tag but I can't compile. I get an error "interpolate_output does not work with COSP". I have interpolate_output = .true. in user_nl_cam, but I'm not setting anything related to COSP. I'm running the same script as I was using with cam6_4_097 and only changed the tag. Do I now have to actively turn COSP off?

@islasimpson, this is a weird interaction in that the default value for the COSP history file is 1 (i.e., h0).

Try setting cosp_histfile_num to something other than 1 in your user_nl_cam.

[cacraigucar]

@gold2718 - Do you think this "weird interaction" is something which warrants opening an issue and a proper default setting? Or would any fix just introduce other problems for COSP users (i.e. they expect COSP to default to h0 and it would no longer do so)

[islasimpson]

Setting cosp_histfile_num to a different number and making sure that history file had interpolate_output = .false. worked to get it to build. It does seem like if this is going to happen by default that will likely cause may people problems.

[gold2718]

@gold2718 - Do you think this "weird interaction" is something which warrants opening an issue and a proper default setting? Or would any fix just introduce other problems for COSP users (i.e. they expect COSP to default to h0 and it would no longer do so)

Short answer: yes

What I would like to do is fix the problem but probably not this week. The new build-namelist block was put in during PR #1359 (cam6_4_109) but I do not see any issue on fixing the underlying problem (interpolated output for multi-dimensional fields such as COSP).

[cacraigucar]

@gold2718 - Thanks for offering to fix this! We are going to be bringing in PR #1355 soon, so if you have the fix ready before we bring it in, we can include it with that (not sure if that tag is going to happen this week or not). If we do move ahead with it before you can provide your fix, we have another opportunity when we bring in #1327 at a later date. If all else fails, we can include it in another tag we make. I'll open an issue to track the error

[islasimpson]

To follow up on the original problem here. I've run a case and then ran a branch using the restart files from that case and it worked fine, so the original problem is fixed. Thank you!