ESMCI / ESMCI/cime

CESM short-term archiver ignores date-stamp on restart files

Open
#4,785 3 comments 0 reactions 0 assignees View on GitHub
Low Priority
Dominant language
Python
Stars
174
Forks
225
Avg merge
1d 16h
Merged PRs (30d)
14

Description

On derecho, I set up a case in `/glade/work/mlevy/codes/CESM/cesm3_0_alpha06d/cases/b.e30a06d.BLT1850.ne30pg3_t232.rpointer_archiving_issue` -- it was fully coupled, and ran for three months writing restarts monthly. I also set `DOUT_S_SAVE_INTERIM_RESTART_FILES=TRUE` to really highlight the issue.

Two things stand out about the `rpointer` files:

1. In the short term archiver, the`rpointer` files from any particular date are archived with that date **and all subsequent dates**. I.e. the February 1 `rpointer` file shows up in `rest/0001-02-01-00000` like it should, but also in `0001-03-01-00000` and `0001-04-01-00000`. The March 1 `rpointer` file is not in `rest/0001-02-01-00000` but is in both `0001-03-01-00000` (good) and `0001-04-01-00000` (bad):

```
$ cd /glade/derecho/scratch/mlevy/archive/b.e30a06d.BLT1850.ne30pg3_t232.rpointer_archiving_issue
$ find . -name rpointer.cam.* | sort
./rest/0001-02-01-00000/rpointer.cam.0001-02-01-00000
./rest/0001-03-01-00000/rpointer.cam.0001-03-01-00000
./rest/0001-04-01-00000/rpointer.cam.0001-02-01-00000
./rest/0001-04-01-00000/rpointer.cam.0001-03-01-00000
./rest/0001-04-01-00000/rpointer.cam.0001-04-01-00000
```

2. After the archiver has run, all 3 months of `rpointer` files remain in the run directory, so if I continue this run they will continue to accumulate in the archiver:

```
$ cd /glade/derecho/scratch/mlevy/b.e30a06d.BLT1850.ne30pg3_t232.rpointer_archiving_issue/run
$ ls -1 rpointer.cam.0001-0*
rpointer.cam.0001-02-01-00000
rpointer.cam.0001-03-01-00000
rpointer.cam.0001-04-01-00000
```

@jedwards4b has suggested we address this issue by updating `config_archiver.xml` in each component, and I'm testing that now. I wasn't sure if there was a fix that could be made at the CIME level to avoid needing to update each component, or if this issue should just be closed without changing any code -- in the latter case, at least this issue can provide a record of how CESM handled this if other modeling centers run into the same problem.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.