File parm/system.conf.zeus

This is the system-specific configuration file for the NOAA Zeus cluster. It should not be used directly. Users should copy it to parm/system.conf and edit the values manually.

This is a configuration override file. This file sets options in the following sections:

• [config] — Sets basic configuration options used by all components.
• [dir] — Configure directory paths.
• [ensda_wrfexe] — Override settings for I/O servers in the ensda initialization jobs.
• [holdvars] — Holdvars variable generation section. This sets variables that are only used to create the holdvars file for ksh. Nothing in this section is ever used by the Python code. That vestigial file is just for compatibility with legacy external workflows, and it will be removed eventually. The external workflows that need it are: HHS - to decide whether things failed graphics scripts - for data paths Note that you must also edit parm/hwrf_holdvars.txt in order for the values in this section to be sent to the holdvars.txt file.
• [hwrf_da_ens] — Configures the hwrf.ensda.DAEnsemble
• [hwrfdata] — HWRF staging area paths. See hwrf.input.DataCatalog for details.
• [runwrf] — This section is read by the script that runs the WRF forecast job at 2km. It is not used by other executions of wrf. See runwrf3km for the 3km configuration.
• [wrfexe] — This section is read by all scripts that run the wrf or nmm_real.

Section [hwrf_da_ens]

Configures the hwrf.ensda.DAEnsemble

See the main documentation for [hwrf_da_ens] for details.

Options in this section:

• realinit — Point to alternate section for init jobs

[hwrf_da_ens] realinit

Point to alternate section for init jobs

[hwrf_da_ens]
realinit = ensda_wrfexe

Defined in File parm/system.conf.zeus

Section [hwrfdata]

HWRF staging area paths. See hwrf.input.DataCatalog for details.

See the main documentation for [hwrfdata] for details.

Options in this section:

• inputroot — Staging area root directory

[hwrfdata] inputroot

[hwrfdata]
inputroot = /scratch2/portfolios/NCEPDEV/ptmp/Samuel.Trahan/hwrfdata

Defined in File parm/system.conf.zeus

Section [ensda_wrfexe]

Override settings for I/O servers in the ensda initialization jobs.

For some odd reason, the ensda wrfanl crashes if I/O servers are enabled and PNetCDF is used. The workaround is to disable I/O servers.

Options in this section:

• nio_groups — Disable I/O servers (groups=1, per group=0)
• nio_tasks_per_group — Disable I/O servers (groups=1, per group=0)
• nproc_x — Init job WRF processor count in X direction (-1 = automatic)
• nproc_y — Init job WRF processor count in Y direction (-1 = automatic)

[ensda_wrfexe] nio_groups

Disable I/O servers (groups=1, per group=0)

[ensda_wrfexe]
nio_groups = 1

Defined in File parm/system.conf.zeus

[ensda_wrfexe] nio_tasks_per_group

Disable I/O servers (groups=1, per group=0)

[ensda_wrfexe]
nio_tasks_per_group = 0,0,0

Defined in File parm/system.conf.zeus

[ensda_wrfexe] nproc_x

Init job WRF processor count in X direction (-1 = automatic)

[ensda_wrfexe]
nproc_x = -1

Defined in File parm/system.conf.zeus

[ensda_wrfexe] nproc_y

Init job WRF processor count in Y direction (-1 = automatic)

[ensda_wrfexe]
nproc_y = -1

Defined in File parm/system.conf.zeus

Section [runwrf]

This section is read by the script that runs the WRF forecast job at 2km. It is not used by other executions of wrf. See runwrf3km for the 3km configuration.

See the main documentation for [runwrf] for details.

Options in this section:

• wrf_ranks — Number of WRF ranks including compute and I/O
• nio_groups — Number of I/O server groups per domain
• nio_tasks_per_group — Number of I/O servers per group
• poll_servers — Turn on server polling if quilt servers are used (they are not).
• nproc_x — WRF processor count in X direction (-1 = automatic)
• nproc_y — WRF processor count in Y direction (-1 = automatic)

[runwrf] wrf_ranks

Number of WRF ranks including compute and I/O

[runwrf]
wrf_ranks = 460

Defined in File parm/system.conf.zeus

[runwrf] nio_groups

Number of I/O server groups per domain

[runwrf]
nio_groups = 1

Defined in File parm/system.conf.zeus

[runwrf] nio_tasks_per_group

Number of I/O servers per group

[runwrf]
nio_tasks_per_group = 4,4,4

Defined in File parm/system.conf.zeus

[runwrf] poll_servers

Turn on server polling if quilt servers are used (they are not).

[runwrf]
poll_servers = yes

Defined in File parm/system.conf.zeus

[runwrf] nproc_x

WRF processor count in X direction (-1 = automatic)

[runwrf]
nproc_x = 16

Defined in File parm/system.conf.zeus

[runwrf] nproc_y

WRF processor count in Y direction (-1 = automatic)

[runwrf]
nproc_y = 28

Defined in File parm/system.conf.zeus

Section [holdvars]

Holdvars variable generation section. This sets variables that are only used to create the holdvars file for ksh. Nothing in this section is ever used by the Python code. That vestigial file is just for compatibility with legacy external workflows, and it will be removed eventually. The external workflows that need it are: HHS - to decide whether things failed graphics scripts - for data paths Note that you must also edit parm/hwrf_holdvars.txt in order for the values in this section to be sent to the holdvars.txt file.

See the main documentation for [holdvars] for details.

Options in this section:

• WHERE_AM_I — Which cluster (for setting up environment)
• WHICH_JET — Which part of Jet? None; we are on Zeus.
• NPROCS_A_NOIO — Unused.
• NPROCS_C — Unused.
• NPROCS_O — Unused.

[holdvars] WHERE_AM_I

Which cluster (for setting up environment)

[holdvars]
WHERE_AM_I = zeus

Defined in File parm/system.conf.zeus

[holdvars] WHICH_JET

Which part of Jet? None; we are on Zeus.

[holdvars]
WHICH_JET = none

Defined in File parm/system.conf.zeus

[holdvars] NPROCS_A_NOIO

Unused.

[holdvars]
NPROCS_A_NOIO = 184

Defined in File parm/system.conf.zeus

[holdvars] NPROCS_C

Unused.

[holdvars]
NPROCS_C = 1

Defined in File parm/system.conf.zeus

[holdvars] NPROCS_O

Unused.

[holdvars]
NPROCS_O = 9

Defined in File parm/system.conf.zeus

Section [wrfexe]

This section is read by all scripts that run the wrf or nmm_real.

See the main documentation for [wrfexe] for details.

Options in this section:

• nio_groups — Number of WRF I/O server groups per domain in init jobs
• nio_tasks_per_group — Number of I/O servers per group in WRF in init jobs.
• poll_servers — Turn on server polling if quilt servers are used (They are not.)
• nproc_x — Init job WRF processor count in X direction (-1 = automatic)
• nproc_y — Init job WRF processor count in Y direction (-1 = automatic)

[wrfexe] nio_groups

Number of WRF I/O server groups per domain in init jobs

[wrfexe]
nio_groups = 1

Defined in File parm/system.conf.zeus

[wrfexe] nio_tasks_per_group

Number of I/O servers per group in WRF in init jobs.

[wrfexe]
nio_tasks_per_group = 12,12,12

Defined in File parm/system.conf.zeus

[wrfexe] poll_servers

Turn on server polling if quilt servers are used (They are not.)

[wrfexe]
poll_servers = yes

Defined in File parm/system.conf.zeus

[wrfexe] nproc_x

Init job WRF processor count in X direction (-1 = automatic)

[wrfexe]
nproc_x = 4

Defined in File parm/system.conf.zeus

[wrfexe] nproc_y

Init job WRF processor count in Y direction (-1 = automatic)

[wrfexe]
nproc_y = 12

Defined in File parm/system.conf.zeus

Section [config]

Sets basic configuration options used by all components.

See the main documentation for [config] for details.

Options in this section:

• input_sources — Specify input sources for HISTORY mode.
• fcst_catalog — Specify the hwrf.input.DataCatalog for FORECAST mode runs.
• post_runs_wrfcopier — Workaround for SGI MPT bug: copy WRF files in the post job
• archive — Default archive path

[config] input_sources

Specify input sources for HISTORY mode.

[config]
input_sources = zeus_sources_{GFSVER}

Specifies input sources for use by hwrf.input.InputSources. In a HISTORY mode run, this will be used in the scripts.exhwrf_input job to pull data.

Defined in File parm/system.conf.zeus

[config] fcst_catalog

Specify the hwrf.input.DataCatalog for FORECAST mode runs.

[config]
fcst_catalog = zeus_fcst_{GFSVER}

Defined in File parm/system.conf.zeus

[config] post_runs_wrfcopier

Workaround for SGI MPT bug: copy WRF files in the post job

[config]
post_runs_wrfcopier = yes

This is a workaround for an SGI MPT bug. MPT cannot handle nested MPI_Init calls, and the Zeus version of ncks is compiled with MPI support. That means the products job cannot run the ncks program to convert NetCDF 3 files to compressed NetCDF 4. As a workaround, we do that in the post job instead.

Defined in File parm/system.conf.zeus

[config] archive

Default archive path

[config]
archive = hpss:/NCEPDEV/emc-hwrf/1year/{ENV[USER]}/{SUBEXPT}/{out_prefix}.tar

Defined in File parm/system.conf.zeus

Section [dir]

Configure directory paths.

See the main documentation for [dir] for details.

Options in this section:

• CDSAVE — Save directory. Make sure you edit this.
• CDNOSCRUB — Non-scrubbed directory for track files, etc. Make sure you edit this.
• CDSCRUB — Scrubbed directory for large work files. Make sure you edit this.
• syndat — Syndat directory for finding which cycles to run

[dir] CDSAVE

Save directory. Make sure you edit this.

[dir]
CDSAVE = /scratch1/portfolios/NCEPDEV/hwrf/save/{ENV[USER]}

Defined in File parm/system.conf.zeus

[dir] CDNOSCRUB

Non-scrubbed directory for track files, etc. Make sure you edit this.

[dir]
CDNOSCRUB = /scratch1/portfolios/NCEPDEV/hwrf/noscrub/{ENV[USER]}

Defined in File parm/system.conf.zeus

[dir] CDSCRUB

Scrubbed directory for large work files. Make sure you edit this.

[dir]
CDSCRUB = /scratch2/portfolios/NCEPDEV/ptmp/{ENV[USER]}

Defined in File parm/system.conf.zeus

[dir] syndat

Syndat directory for finding which cycles to run

[dir]
syndat = /scratch1/portfolios/NCEPDEV/hwrf/noscrub/input/SYNDAT-PLUS

Defined in File parm/system.conf.zeus