Issue with ./case.build

[jzweifel]

Hi all,

I'm working on an experiment where I have ran a control simulation 50 years and am now trying to go back to implement a forcing related to sulfur dioxide emissions.

To do this, I created 20 cloned cases from my control simulation and am now trying to run them from June from year 0031-0050 respectively. Before I try to write a script to automate this process, I wanted to try and do this manually. Here are the runtime variables I have set in a case called clone_1.

Results in group run_begin_stop_restart
CONTINUE_RUN: TRUE
ESP_RUN_ON_PAUSE: FALSE
RESUBMIT_SETS_CONTINUE_RUN: TRUE
RUN_REFCASE: control_F2000
RUN_REFDATE: 0031-06-01
RUN_REFDIR: /glade/derecho/scratch/jzweifel/archive/control_F2000/rest/next_20_years/june_rests/0031-06-01-00000
RUN_REFTOD: 00000
RUN_STARTDATE: 0031-06-01
RUN_TYPE: hyrbid

Ideally I would want this clone_1 model to simulate starting from the base conditions of control_F2000 as they were in 0031-06-01. Would this combination of variables achieve that goal?

I thought they would and I tried to then use ./case.build on clone_1 and am getting the following feedback:

jzweifel@derecho3:/glade/work/jzweifel/cases/clone_1> ./case.build Building case in directory /glade/work/jzweifel/cases/clone_1 sharedlib_only is False model_only is False File /glade/work/jzweifel/cases/clone_1/LockedFiles/env_build.xml has been modified Generating component namelists as part of build Creating component namelists Calling /glade/work/jzweifel/cases/derecho_cesm_sandbox/cime/../components/cam//cime_config/buildnml ...calling cam buildcpp to set build time options

This ./case.build is taking much much longer than before and hasn't moved on from this message in hours. I get the sense that something must be incorrect. Would anyone be able to help me with this?

Thanks so much for all the help in the past/present/future!

Jack

 

[jzweifel]

Hi all, something is definitely incorrect with my clone_1 case that I am trying run, yesterday before leaving my office I tried running: nohup ./case.build & to try and let my case build while I was away from my/cesm machines. The case still did not finish building and was stuck at the above.

Unfortunately I also got this email just a few minutes ago from UCAR:

A violation of the usage policy by jzweifel (Jack Zweifel,UNIVERSITY OF WISCONSIN AT MADISON) on derecho4 was automatically detected starting at 09:40 on 02/29.

I immediately killed that case.build in response!

I'm still really not sure why this is happening, but here are a few of my thoughts:

a) my xml variables are not configured correctly to get my clone_1 to initialize from 0031-06-01 from control_F2000 and run for 3 more months now with the sulfur dioxide forcing I inputed into its user_nl_cam.

b) Also in my user_nl_cam I tried to add more output variables using the fincl method I found... maybe this is somehow messing something up.

c) when I used to do work on Cheyenne and wanted to build a case I would call qcmd -- ./case.build, this always worked well, but when I tried to build a case my first time on Derecho I tried using that method and couldn't progress forward. My thought was just that Derecho being a new machine maybe didn't require the qcmd preamble, so I had just been using ./case.build directly, it had worked just fine for my other cases. Maybe I was incorrect here and I should still be using qcmd first. I just tried it now and got this feedback:

jzweifel@derecho4:/glade/work/jzweifel/cases/clone_1> qcmd -- ./case.build Error: no project provided as argument or via PBS_ACCOUNT

I'm a little confused on that error and don't know how to get around it. And even if I did, I still have a suspicion that something else is incorrectly configured in my case.

Any help would be appreciated, I think once I get this case figured out and running smoothly I'll be good to move on to repeating the process with my other clones!

Thanks again!

Jack