Skip to content

Job Manifests

JobManifest

Bases: ExtendedBaseModel

Model for parsing job manifests: describes the expected layout of a job manifest and provides methods for returning job definitions that can be used to initialize a slurmjobs job instance (currently, SlurmJob)

Add validator to ensure that depencies for a job do not list the job itself

as a dependency.

Attributes:

Name Type Description
job_options Optional[List[JobOptions]]

section of YML job manifest for defining global job options as a list. Global job options are tagged using anchors and referenced within a given job's job definition, under the job field. All job options that

jobs Optional[List[Job]]

list of instances of JobDefinition
Source code in catena/models/job_manifest.py 
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
class JobManifest(ExtendedBaseModel):
    """
    Model for parsing job manifests: describes the expected layout of a job manifest
    and provides methods for returning job definitions that can be used to initialize
    a `slurmjobs` job instance (currently, SlurmJob)

    TODO: Add validator to ensure that depencies for a job do not list the job itself
          as a dependency.

    Attributes:
        job_options: section of YML job manifest for defining global job options as a list.
            Global job options are tagged using anchors and referenced within a given job's
            job definition, under the `job` field. All job options that 

        jobs:  list of instances of JobDefinition 

    """
    cluster_profile: str
    catena_config: Optional[str] = str(Path().home() / '.catena/conf.yml')
    version: Optional[str] = "1.0"
    job_options: Optional[List[JobOptions]]
    jobs: Optional[List[Job]]

    class Config:
        """
        `JobManifest` model `Config` sub-class: can be accessed internally
        as `self.Config` or `JobManifest.Config`. The config class of a Pydantic model is also
        passed as a kwarg to validators and acts as a useful means of 
        storing external variables that are accesible to field 
        validators.

        Attributes:
            ext_opts: extra job options that are separate from SLURM job options.
                This difference is abstracted to the user, but is managed
                internally.
        """
        ext_opts: list = ['env_modules', 
                          'job_script',
                          'env_extra',
                          'job_script_args',
                          'command',
                          'dependencies']


    def __filter_ext_opts(self, jobdef: JobOptions, field: str):
        """
        Filter non sbatch options from job (global opts) field. Local
        job opts will always take precedence over those defined globally
        """

        field_val = None
        if (jobdef.job.dict().get(field) is not None and 
            jobdef.dict().get(field) is not None):
            field_val = jobdef.pop(field)
            jobdef.job.pop(field)

        elif jobdef.job.dict().get(field) is not None: #and
              field_val = jobdef.job.pop(field)

        else:
            field_val = jobdef.pop(field)

        return jobdef, field_val


    def expand_jobs(self):
        """
        Return list of job definitions processed from a initialized instance
        of `self` or `JobManifest`. The fields defined within this model can
        be passed to an instance of a `slurmjobs` Job. For example, the 
        [SLURMRESTJob](../jobs/slurmrestjob.md) object.
        """
        jobs = []
        filtered_opts = {}

        # iterate over all job definitions
        for jobblock in self.jobs:

            for jobname, jobdef in jobblock.__root__.items():

                # create named tuple for storing unset/set options and there default values
                Options = namedtuple('Options', 'name default')     

                # collect set job options on local and global levels
                local_opts_set = jobdef.__fields_set__
                global_opts_set = jobdef.__fields_set__
                fields = jobdef.job.__fields__

                # set job name
                jobdef.name = jobname

                # create copy of job excluding env_modules property
                tmpjob = JobOptions(**{k:v for k,v in jobdef.job.dict().items() 
                                           if k != 'env_modules'})
                tmpjob = JobOptions(**jobdef.job.dict())

                # filter external options
                for optname in self.Config.ext_opts:
                    jobdef, optval = self.__filter_ext_opts(jobdef, optname)
                    filtered_opts[optname] = optval

                # list all set field names and default values for non-required fields      
                setfields = [Options(x, fields[x].default) for x in fields.keys() if x in 
                                global_opts_set]

                for opt in setfields:
                    if opt.name in jobdef.dict():
                        tmpjob[opt.name] = jobdef.pop(opt.name)

                # add ext opts back into job def
                for optname, optval in filtered_opts.items():
                    jobdef[optname] = optval
                    tmpjob.pop(optname)

                # set job properties after filtering
                jobdef['job'] = tmpjob
                jobs.append(jobdef)

        return jobs

Config

JobManifest model Config sub-class: can be accessed internally as self.Config or JobManifest.Config. The config class of a Pydantic model is also passed as a kwarg to validators and acts as a useful means of storing external variables that are accesible to field validators.

Attributes:

Name Type Description
ext_opts list

extra job options that are separate from SLURM job options. This difference is abstracted to the user, but is managed internally.
Source code in catena/models/job_manifest.py 
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
class Config:
    """
    `JobManifest` model `Config` sub-class: can be accessed internally
    as `self.Config` or `JobManifest.Config`. The config class of a Pydantic model is also
    passed as a kwarg to validators and acts as a useful means of 
    storing external variables that are accesible to field 
    validators.

    Attributes:
        ext_opts: extra job options that are separate from SLURM job options.
            This difference is abstracted to the user, but is managed
            internally.
    """
    ext_opts: list = ['env_modules', 
                      'job_script',
                      'env_extra',
                      'job_script_args',
                      'command',
                      'dependencies']

__filter_ext_opts(jobdef, field)

Filter non sbatch options from job (global opts) field. Local job opts will always take precedence over those defined globally

Source code in catena/models/job_manifest.py 
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
def __filter_ext_opts(self, jobdef: JobOptions, field: str):
    """
    Filter non sbatch options from job (global opts) field. Local
    job opts will always take precedence over those defined globally
    """

    field_val = None
    if (jobdef.job.dict().get(field) is not None and 
        jobdef.dict().get(field) is not None):
        field_val = jobdef.pop(field)
        jobdef.job.pop(field)

    elif jobdef.job.dict().get(field) is not None: #and
          field_val = jobdef.job.pop(field)

    else:
        field_val = jobdef.pop(field)

    return jobdef, field_val

expand_jobs()

Return list of job definitions processed from a initialized instance of self or JobManifest. The fields defined within this model can be passed to an instance of a slurmjobs Job. For example, the SLURMRESTJob object.

Source code in catena/models/job_manifest.py 
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
def expand_jobs(self):
    """
    Return list of job definitions processed from a initialized instance
    of `self` or `JobManifest`. The fields defined within this model can
    be passed to an instance of a `slurmjobs` Job. For example, the 
    [SLURMRESTJob](../jobs/slurmrestjob.md) object.
    """
    jobs = []
    filtered_opts = {}

    # iterate over all job definitions
    for jobblock in self.jobs:

        for jobname, jobdef in jobblock.__root__.items():

            # create named tuple for storing unset/set options and there default values
            Options = namedtuple('Options', 'name default')     

            # collect set job options on local and global levels
            local_opts_set = jobdef.__fields_set__
            global_opts_set = jobdef.__fields_set__
            fields = jobdef.job.__fields__

            # set job name
            jobdef.name = jobname

            # create copy of job excluding env_modules property
            tmpjob = JobOptions(**{k:v for k,v in jobdef.job.dict().items() 
                                       if k != 'env_modules'})
            tmpjob = JobOptions(**jobdef.job.dict())

            # filter external options
            for optname in self.Config.ext_opts:
                jobdef, optval = self.__filter_ext_opts(jobdef, optname)
                filtered_opts[optname] = optval

            # list all set field names and default values for non-required fields      
            setfields = [Options(x, fields[x].default) for x in fields.keys() if x in 
                            global_opts_set]

            for opt in setfields:
                if opt.name in jobdef.dict():
                    tmpjob[opt.name] = jobdef.pop(opt.name)

            # add ext opts back into job def
            for optname, optval in filtered_opts.items():
                jobdef[optname] = optval
                tmpjob.pop(optname)

            # set job properties after filtering
            jobdef['job'] = tmpjob
            jobs.append(jobdef)

    return jobs

JobOptions

Bases: SlurmSubmit

Job options schema for job manifest. Extends SlurmSubmit which provides attributes corresponding to sbatch options in SLURM.

Attributes:

Name Type Description
name Optional[str]

job name as it will appear in SLURM queue (the root key of a given Job is usually set internally as the job name when parsing a manifest)
env_modules Optional[List[str]]

list module names that should be loaded when submitting the job to a SLURM scheduler and made available to the environment running the job. For example, ['anaconda3/2021.11', 'matlab/96']
env_extra Optional[Dict[str, Any]]

dictionary of extra environment variables to export to the environment running the job submitted to a SLURM scheduler. For example, {'MYVAR': '/path/to/stuff'}. See how to append, prepend, or replace
job_script Optional[str]

path to the script, either relative or absolute. The ~ is also excepted and converted to the appropriate absolute path for the current users directory. If using relative paths, be sure to understand where the paths should be relative to, to appropriately locate your scripts.
job_script_args Optional[List[str]]

list of arguments to pass to the job_script when it is executed. For example, job_script_args=['--version']
command Optional[str]

custom command to use when running job_script. If defined, the the script provided will be run as an executable with this command prepended. For example, command='python -m' would result in the job_script being called as: python -m <job_script>. Generally, the default should give the right result.
Source code in catena/models/job_manifest.py 
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
class JobOptions(SlurmSubmit):
    """
    Job options schema for job manifest. Extends SlurmSubmit which
    provides attributes corresponding to sbatch options in SLURM. 


    Attributes:
        name: job name as it will appear in SLURM queue (the __root__ key of a
            given Job is usually set internally as the job name when parsing a manifest)

        env_modules: list module names that should be loaded when submitting the job
            to a SLURM scheduler and made available to the environment running the job. 
            For example,  `['anaconda3/2021.11', 'matlab/96']`

        env_extra: dictionary of extra environment variables to export to the
            environment running the job submitted to a SLURM scheduler. For
            example, `{'MYVAR': '/path/to/stuff'}`. See how to 
            [append, prepend, or replace](../examples/example1.md#env_extra_feats)

        job_script: path to the script, either relative or absolute. The `~` is
            also excepted and converted to the appropriate absolute path for the 
            current users directory. ***If*** using relative paths, be sure to
            understand where the paths should be relative to, to appropriately
            locate your scripts.

        job_script_args: list of arguments to pass to the job_script when
            it is executed. For example, `job_script_args=['--version']`

        command: custom command to use when running `job_script`. If defined, the
            the script provided will be run as an executable with this command
            prepended. For example, command='python -m' would result in the
            `job_script` being called as: `python -m <job_script>`. Generally,
            the default should give the right result.
    """
    name: Optional[str]
    env_modules: Optional[List[str]] = None
    env_extra: Optional[Dict[str, Any]] = None
    job_script: Optional[str]
    job_script_args: Optional[List[str]] = None
    command: Optional[str]
    dependencies: Optional[Dict[DependencyType, Union[str, List[str]]]]

    @validator('job_script')
    def expand_home_shortcut(cls, v):
        if v is not None:
            if str(v.startswith('~')):
                ppath = Path(v)
                return str(ppath.expanduser())
            else:
                return v
        else:
            return v

JobDefinition

Bases: JobOptions

Local job definition that extends JobOptions as well as provides a field, job, for referencing global JobOptions (using YML anchors/aliases)

Attributes:

Name Type Description
job Optional[JobOptions]

global job options parsed from YML anchors/aliases. Ultimately local and global job options are combined when defining a given Job to be submitted to a SLURM cluster, with local job options taking precendence over global
Source code in catena/models/job_manifest.py 
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
class JobDefinition(JobOptions):
    """
    Local job definition that extends JobOptions as well as provides
    a field, job, for referencing global JobOptions (using YML
    anchors/aliases)

    Attributes:

        job: global job options parsed from YML anchors/aliases. Ultimately local
            and global job options are combined when defining a given Job to be 
            submitted to a SLURM cluster, with local job options taking precendence
            over global

    """
    job: Optional[JobOptions]