Need capability to run multiple forecast segments #2274
Assignees
Labels
feature
New feature or request
Comments
7 tasks
10 tasks
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Jul 25, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
7 tasks
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Jul 25, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Jul 26, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Jul 30, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Jul 30, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Aug 2, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Aug 5, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Aug 6, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Aug 7, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
WalterKolczynski-NOAA
added a commit
to WalterKolczynski-NOAA/global-workflow
that referenced
this issue
Aug 8, 2024
Adds the ability to run a forecast in segments instead of all at
once. To accomplish this, the `FHMIN_GFS` and `FHMAX_GFS` settings
have been replaced as user-setable variables in favor of
`FCST_SEGMENTS_STR_GFS`, a comma-separated list of the segment
boundaries (thus there will be one more than the number of segments).
For a traditional single-segment forecast, this would just be set to
`"${FHMIN_GFS},${FHMAX_GFS}"`.
The comma-separated list had to be used instead of a bash array as
the variable must be exported in order for the rocoto generator to
see it, and arrays cannot be exported from shell. Capabilty to parse
these into python lists was added to wxflow in an accompanying PR.
To accomodate the new segment metatasks that must be run serially,
the capability of `create_task()` was expanded to allow a dictionary
key of `is_serial`, which controls whether a metatask is parallel or
serial using pre-existing capability in rocoto. The default when not
given is parallel (i.e. most metatasks).
Resolves NOAA-EMC#2274
Refs NOAA-EMC/wxflow#39
DavidHuber-NOAA
pushed a commit
that referenced
this issue
Aug 12, 2024
Adds the ability to run a forecast in segments instead of all at once. To accomplish this, a new local `checkpnts` variable is introduced to `config.base` to contain a comma-separated list of intermediate stopping points for the forecast. This is combined with `FHMIN_GFS` and `FHMAX_GFS` to create a comma-separated string `FCST_SEGMENTS` with all the start/end points that is used by `config.fcst` and rocoto workflow. Capability to parse these into python lists was added to wxflow in an accompanying PR. If `checkpnts` is an empty string, this will result in a single-segment forecast. To accommodate the new segment metatasks that must be run serially, the capability of `create_task()` was expanded to allow a dictionary key of `is_serial`, which controls whether a metatask is parallel or serial using pre-existing capability in rocoto. The default when not given is parallel (i.e. most metatasks). Resolves #2274 Refs NOAA-EMC/wxflow#39 Refs NOAA-EMC/wxflow#40
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
What new functionality do you need?
Longer forecasts (such as those for GEFS S2S and SFS) are too long to run in a single job within wallclock limits and/or fit in different, heterogenous chunks in operations.
Depends on #2273
Depends on #2674
What are the requirements for the new functionality?
The ability to run a forecast in multiple segments, which each segment continuing from the end of the previous one. At least two are needed, but allowing an arbitrary number is best (and may be necessary for SFS).
Stretch scope: allow different resources for each segment. This can be handled as a separate issue unless it makes sense to incorporate into the design here.
Acceptance Criteria
Suggest a solution (optional)
No response
The text was updated successfully, but these errors were encountered: