class UpdateDagRunsRequestSchema(ApiStrictSchema):
dag_runs = fields.Nested(DagRunSchema, many=True)
task_instances = fields.Nested(TaskInstanceSchema, many=True)
last_seen_log_id = fields.Integer(allow_none=True)
airflow_export_meta = fields.Nested(AirflowExportMetaSchema, required=False)
error_message = fields.String(required=False, allow_none=True)
syncer_type = fields.String(allow_none=True)
class DatasetsReportSchema(ApiStrictSchema):
sync_event_uid = fields.UUID()
monitor_state = fields.Nested(DatasourceMonitorStateSchema)
source_type = fields.String(allow_none=True) # bigquery / snowflake / etc
syncer_type = fields.String(allow_none=True)
sync_event_timestamp = fields.DateTime(required=False, allow_none=True)
datasets = fields.Nested(SyncedDatasetSchema, many=True)
@post_load
def make_object(self, data):
return DatasetsReport(**data)
class SyncedTransactionsReportSchema(ApiStrictSchema):
sync_event_uid = fields.UUID()
sync_event_timestamp = fields.DateTime()
source_type = fields.String(allow_none=True) # bigquery / snowflake / etc
syncer_type = fields.String(allow_none=True)
monitor_state = fields.Nested(DatasourceMonitorStateSchema)
transactions = fields.Nested(SyncedTransactionSchema, many=True)
@post_load
def make_object(self, data):
return SyncedTransactionsReport(**data)
class TaskRunAttemptSchema(ApiStrictSchema):
task_run_uid = fields.UUID()
task_run_attempt_uid = fields.UUID()
state = EnumField(TaskRunState, allow_none=True)
timestamp = fields.DateTime(allow_none=True)
first_error = fields.Nested(ErrorInfoSchema, allow_none=True)
latest_error = fields.Nested(ErrorInfoSchema, allow_none=True)
attempt_number = fields.Number(allow_none=True)
source = fields.Str(allow_none=True)
start_date = fields.DateTime(allow_none=True)
end_date = fields.DateTime(allow_none=True)
external_links_dict = fields.Dict(allow_none=True)
class AirflowServerInfoSchema(ApiStrictSchema):
airflow_version = fields.String(allow_none=True)
airflow_export_version = fields.String(allow_none=True)
airflow_monitor_version = fields.String(allow_none=True)
last_sync_time = fields.DateTime(allow_none=True)
monitor_error_message = fields.String(allow_none=True)
synced_to = fields.DateTime(allow_none=True)
api_mode = fields.String(allow_none=True)
sync_interval = fields.Integer(allow_none=True)
is_sync_enabled = fields.Boolean(allow_none=True)
system_alert_definitions = fields.Dict()
fetcher = fields.String(allow_none=True)
composer_client_id = fields.String(allow_none=True)
dag_ids = fields.String(allow_none=True)
base_url = fields.String()
external_url = fields.String(allow_none=True)
source_instance_uid = fields.String(allow_none=True)
tracking_source_uid = fields.String()
airflow_instance_uid = fields.String(
allow_none=True) # TODO_API: deprecate
name = fields.String(allow_none=True)
env = fields.String(allow_none=True)
monitor_status = fields.String(allow_none=True)
monitor_config = fields.Nested(MonitorConfigSchema, allow_none=True)
airflow_environment = fields.String(allow_none=True)
last_seen_dag_run_id = fields.Integer(allow_none=True)
last_seen_log_id = fields.Integer(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return AirflowServerInfo(**data)
class ScheduledJobSchemaV2(Schema):
class Meta:
strict = True
uid = fields.Str(attribute="DbndScheduledJob.uid", allow_none=True)
name = fields.Str(attribute="DbndScheduledJob.name", required=True)
cmd = fields.Str(attribute="DbndScheduledJob.cmd", required=True)
schedule_interval = fields.Str(
attribute="DbndScheduledJob.schedule_interval", required=True)
start_date = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.start_date",
format="iso")
end_date = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.end_date",
format="iso")
readable_schedule_interval = fields.Str(
attribute="DbndScheduledJob.readable_schedule_interval",
allow_none=True)
scheduled_interval_in_seconds = fields.Integer(
attribute="DbndScheduledJob.scheduled_interval_in_seconds",
allow_none=True)
catchup = fields.Boolean(allow_none=True,
attribute="DbndScheduledJob.catchup")
depends_on_past = fields.Boolean(
allow_none=True, attribute="DbndScheduledJob.depends_on_past")
retries = fields.Int(allow_none=True, attribute="DbndScheduledJob.retries")
active = fields.Boolean(allow_none=True,
attribute="DbndScheduledJob.active")
create_user = fields.Str(allow_none=True,
attribute="DbndScheduledJob.create_user")
create_time = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.create_time")
update_user = fields.Str(allow_none=True,
attribute="DbndScheduledJob.update_user")
update_time = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.update_time")
from_file = fields.Boolean(allow_none=True,
attribute="DbndScheduledJob.from_file")
deleted_from_file = fields.Boolean(
allow_none=True, attribute="DbndScheduledJob.deleted_from_file")
next_job_date = fields.DateTime(attribute="DbndScheduledJob.next_job_date",
allow_none=True)
alerts = fields.List(
fields.Nested(AlertEventSchema),
attribute="DbndScheduledJob.alerts",
allow_none=True,
)
job_name = fields.Str(dump_only=True,
attribute="DbndScheduledJob.job_name")
last_run_uid = fields.UUID(dump_only=True)
last_run_job = fields.Str(dump_only=True)
last_job_date = fields.DateTime(dump_only=True)
last_run_state = fields.Str(dump_only=True)
is_airflow_synced = fields.Bool(dump_only=True)
list_order = fields.Integer(attribute="DbndScheduledJob.list_order",
allow_none=True)
validation_errors = fields.Str(
allow_none=True, attribute="DbndScheduledJob.validation_errors")
class SyncedTransactionSchema(ApiStrictSchema):
datasource_transaction_id = fields.String()
created_date = fields.DateTime()
started_date = fields.DateTime()
ended_date = fields.DateTime()
write_operation = fields.Nested(SyncedTransactionOperationSchema)
read_operations = fields.Nested(
SyncedTransactionOperationSchema, many=True, allow_none=True
)
query_string = fields.String(allow_none=True)
data_schema = fields.Nested(StructuredDataSchema, allow_none=True)
@post_load
def make_object(self, data):
return SyncedTransaction(**data)
class InitRunArgsSchema(ApiObjectSchema):
run_uid = fields.UUID()
root_run_uid = fields.UUID()
driver_task_uid = fields.UUID(allow_none=True)
task_run_env = fields.Nested(TaskRunEnvInfoSchema)
task_runs_info = fields.Nested(TaskRunsInfoSchema)
new_run_info = fields.Nested(RunInfoSchema, allow_none=True)
scheduled_run_info = fields.Nested(ScheduledRunInfoSchema, allow_none=True)
update_existing = fields.Boolean()
source = fields.Str(allow_none=True)
@post_load
def make_init_run_args(self, data, **kwargs):
return InitRunArgs(**data)
class LogMetricsSchema(_ApiCallSchema):
metrics_info = fields.Nested(LogMetricSchema, many=True)
@post_load
def make_object(self, data, **kwargs):
data["metrics_info"] = [
LogMetricArgs(**m) for m in data["metrics_info"]
]
class RunInfoSchema(ApiStrictSchema):
root_run_uid = fields.UUID()
run_uid = fields.UUID()
job_name = fields.String()
project_name = fields.String(allow_none=True)
user = fields.String()
name = fields.String()
description = fields.String(allow_none=True)
state = EnumField(RunState)
start_time = fields.DateTime()
end_time = fields.DateTime(allow_none=True)
# deprecate
dag_id = fields.String()
cmd_name = fields.String(allow_none=True)
execution_date = fields.DateTime()
# move to task
target_date = fields.Date(allow_none=True)
version = fields.String(allow_none=True)
driver_name = fields.String()
is_archived = fields.Boolean()
env_name = fields.String(allow_none=True)
cloud_type = fields.String()
trigger = fields.String()
task_executor = fields.String(allow_none=True)
root_run = fields.Nested(RootRunInfoSchema)
scheduled_run = fields.Nested(ScheduledRunInfoSchema, allow_none=True)
sends_heartbeat = fields.Boolean(default=False, allow_none=True)
scheduled_job_name = fields.String(allow_none=True)
scheduled_date = fields.DateTime(allow_none=True)
external_links = fields.Dict(allow_none=True)
@post_load
def make_run_info(self, data, **kwargs):
return _as_dotted_dict(**data)
class TaskRunsInfoSchema(ApiObjectSchema):
run_uid = fields.UUID()
root_run_uid = fields.UUID()
task_run_env_uid = fields.UUID()
parent_child_map = fields.List(fields.List(fields.UUID()))
task_runs = fields.Nested(TaskRunInfoSchema,
many=True,
exclude=("task_signature_source", ))
upstreams_map = fields.List(fields.List(fields.UUID()))
dynamic_task_run_update = fields.Boolean()
targets = fields.Nested(TargetInfoSchema, many=True)
task_definitions = fields.Nested(TaskDefinitionInfoSchema, many=True)
af_context = fields.Nested(AirflowTaskContextSchema, allow_none=True)
@post_load
def make_run_info(self, data, **kwargs):
return _as_dotted_dict(**data)
class SyncedDatasetMetadataSchema(ApiStrictSchema):
num_bytes = fields.Integer()
records = fields.Integer()
schema = fields.Nested(StructuredDataSchema)
# TODO: ADD -> Preview
@post_load
def make_object(self, data):
return SyncedDatasetMetadata(**data)
class SyncedDatasetSchema(ApiStrictSchema):
uri = fields.String() # {storage_type://region/project_id/scheme_name/table_name}
created_date = fields.DateTime()
last_modified_date = fields.DateTime(allow_none=True)
metadata = fields.Nested(SyncedDatasetMetadataSchema)
uid = fields.UUID(allow_none=True)
@post_load
def make_object(self, data):
return SyncedDataset(**data)
class TaskRunAttemptUpdateArgsSchema(ApiObjectSchema):
task_run_uid = fields.UUID()
task_run_attempt_uid = fields.UUID()
state = EnumField(TaskRunState, allow_none=True)
timestamp = fields.DateTime(allow_none=True)
error = fields.Nested(ErrorInfoSchema, allow_none=True)
attempt_number = fields.Number(allow_none=True)
source = fields.Str(allow_none=True)
start_date = fields.DateTime(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return TaskRunAttemptUpdateArgs(**data)
class TaskRunInfoSchema(ApiObjectSchema):
task_run_uid = fields.UUID()
task_run_attempt_uid = fields.UUID()
task_definition_uid = fields.UUID()
run_uid = fields.UUID()
task_id = fields.String()
task_signature = fields.String()
task_signature_source = fields.String()
task_af_id = fields.String()
execution_date = fields.DateTime()
name = fields.String()
env = fields.String()
command_line = fields.String()
functional_call = fields.String()
has_downstreams = fields.Boolean()
has_upstreams = fields.Boolean()
is_reused = fields.Boolean()
is_dynamic = fields.Boolean()
is_system = fields.Boolean()
is_skipped = fields.Boolean()
is_root = fields.Boolean()
output_signature = fields.String()
state = EnumField(TaskRunState)
target_date = fields.Date(allow_none=True)
log_local = fields.String(allow_none=True)
log_remote = fields.String(allow_none=True)
version = fields.String()
task_run_params = fields.Nested(TaskRunParamSchema, many=True)
external_links = fields.Dict(allow_none=True)
@post_load
def make_task_run(self, data, **kwargs):
return TaskRunInfo(**data)
class TaskDefinitionInfoSchema(ApiObjectSchema):
task_definition_uid = fields.UUID()
name = fields.String()
class_version = fields.String()
family = fields.String()
module_source = fields.String(allow_none=True)
module_source_hash = fields.String(allow_none=True)
source = fields.String(allow_none=True)
source_hash = fields.String(allow_none=True)
type = fields.String()
task_param_definitions = fields.Nested(TaskDefinitionParamSchema, many=True)
@post_load
def make_task_definition(self, data, **kwargs):
return TaskDefinitionInfo(**data)
class DagSchema(ApiStrictSchema):
description = fields.String()
root_task_ids = fields.List(fields.String())
tasks = fields.Nested(TaskSchema, many=True)
owner = fields.String()
dag_id = fields.String()
schedule_interval = fields.String()
catchup = fields.Boolean(allow_none=True)
start_date = fields.DateTime(allow_none=True)
end_date = fields.DateTime(allow_none=True)
is_committed = fields.Boolean()
git_commit = fields.String()
dag_folder = fields.String()
hostname = fields.String()
source_code = fields.String(allow_none=True)
module_source_hash = fields.String(allow_none=True)
is_subdag = fields.Boolean()
tags = fields.List(fields.String(), allow_none=True)
task_type = fields.String()
task_args = fields.Dict()
is_active = fields.Boolean(allow_none=True)
is_paused = fields.Boolean(allow_none=True)
class UpdateTaskRunAttemptsSchema(_ApiCallSchema):
task_run_attempt_updates = fields.Nested(TaskRunAttemptUpdateArgsSchema, many=True)
class AddTaskRunsSchema(_ApiCallSchema):
task_runs_info = fields.Nested(TaskRunsInfoSchema)
source = EnumField(UpdateSource, allow_none=True)
class InitRunSchema(_ApiCallSchema):
init_args = fields.Nested(InitRunArgsSchema)
class LogMetricSchema(_ApiCallSchema):
task_run_attempt_uid = fields.UUID(required=True)
target_meta_uid = fields.UUID(allow_none=True)
metric = fields.Nested(MetricSchema, allow_none=True)
source = fields.String(allow_none=True)
class AddTaskRunsSchema(_ApiCallSchema):
task_runs_info = fields.Nested(TaskRunsInfoSchema)
class AirflowTaskInfosSchema(_ApiCallSchema):
airflow_task_infos = fields.Nested(AirflowTaskInfoSchema, many=True)
source = fields.String()
base_url = fields.String()
class LogDataframeHistogramListSchema(_ApiCallSchema):
histograms_info = fields.Nested(LogDataframeHistogramSchema)
class LogTargetsSchema(_ApiCallSchema):
targets_info = fields.Nested(LogTargetSchema, many=True)
class AirflowExportMetaSchema(ApiStrictSchema):
airflow_version = fields.String()
plugin_version = fields.String()
request_args = fields.Dict()
metrics = fields.Nested(MetricsSchema)
class ScheduledJobArgsSchema(_ApiCallSchema):
scheduled_job_args = fields.Nested(ScheduledJobInfoSchema)
update_existing = fields.Boolean(default=False)
class AlertDefsSchema(ApiStrictSchema):
severity = fields.Str(required=True)
type = fields.Str(required=True)
user_metric = fields.Str()
operator = fields.Str()
is_str_value = fields.Bool()
created_at = fields.DateTime()
scheduled_job_name = fields.Str(attribute="scheduled_job.name")
source_instance_name = fields.Method("get_tracking_source_name")
env = fields.Method("get_tracking_source_env")
# TODO_CORE: API: Deprecate airflow_server_info
airflow_instance_name = fields.Method("get_tracking_source_name")
project_id = fields.Int(attribute="job.project_id")
project_name = fields.Str(attribute="job.project.name")
alert_on_historical_runs = fields.Bool()
group_uid = fields.Str(allow_none=True, load_from="alert_group_uid")
root_group_uid = fields.Str(allow_none=True,
load_from="alert_root_group_uid")
uid = fields.Str(allow_none=True)
value = fields.Str(allow_none=True)
job_id = fields.Int(allow_none=True)
summary = fields.Str(allow_none=True)
job_name = fields.Str(attribute="job.name", allow_none=True)
task_repr = fields.Str(allow_none=True)
task_name = fields.Str(allow_none=True)
custom_name = fields.Str(allow_none=True)
original_uid = fields.Str(allow_none=True)
advanced_json = fields.Str(allow_none=True)
scheduled_job_uid = fields.Str(allow_none=True)
custom_description = fields.Str(allow_none=True)
ml_alert = fields.Nested(MLAlert, allow_none=True)
# Fields for DatasetSlaAlert/DatasetSlaAdvancedAlert alert
# --------------------------------------
seconds_delta = fields.Int(
allow_none=True) # Converts to datetime.timedelta
dataset_partial_name = fields.Str(allow_none=True)
datasets_uids = fields.List(fields.Str(), allow_none=True)
# Fields for OperationColumnStatAdvancedAlert alert
# --------------------------------------
dataset_uid = fields.Str(allow_none=True)
# Operation type (e.g. "read", "write", None=any) to filter stats by
operation_type = fields.Str(allow_none=True)
# Type of MetricRule, found in dbnd_web. Used to build advanced_json
metrics_rules = fields.List(fields.Dict(), allow_none=True)
# Used only used by the UI
affected_datasets = fields.List(fields.Dict(),
allow_none=True,
dump_only=True)
assigned_jobs = fields.Method(serialize="get_assigned_jobs",
dump_only=True)
is_system = fields.Function(lambda alert_def: alert_def.owner == "system",
dump_only=True)
def get_tracking_source_name(self, obj):
return self._get_tracking_source_instance(obj).name
def get_tracking_source_env(self, obj):
return self._get_tracking_source_instance(obj).env
def _get_tracking_source_instance(self, obj):
if obj.job:
return obj.job.tracking_source
return obj.tracking_source
@pre_load
def prepere(self, data: dict, **kwargs):
value = data.get("value", None)
if value is not None:
data["value"] = str(data["value"])
return data
def get_assigned_jobs(self, alert_def):
self_job = (alert_def.job_id, alert_def.job_name)
sub_alerts_jobs = ((sub_alert.job_id, sub_alert.job_name)
for sub_alert in alert_def.sub_alert_definitions)
alert_jobs = chain(sub_alerts_jobs, [self_job])
alert_jobs = set(filter(lambda l: l != (None, None), alert_jobs))
serialized_assigned_jobs = [{
"job_id": job_id,
"job_name": job_name
} for job_id, job_name in alert_jobs]
return serialized_assigned_jobs