Generate BigConfig files for views (#6312)

* Generate BigConfig files for views

* Re-enable monitoring for telemetry.releases
This commit is contained in:
Anna Scholtz 2024-10-10 07:12:54 -07:00 коммит произвёл GitHub
Родитель 1920e189d7
Коммит c0114f4626
Не найден ключ, соответствующий данной подписи
Идентификатор ключа GPG: B5690EEEBB952194
2 изменённых файлов: 142 добавлений и 51 удалений

Просмотреть файл

@ -12,7 +12,15 @@ from bigeye_sdk.client.datawatch_client import datawatch_client_factory
from bigeye_sdk.client.enum import Method from bigeye_sdk.client.enum import Method
from bigeye_sdk.controller.metric_suite_controller import MetricSuiteController from bigeye_sdk.controller.metric_suite_controller import MetricSuiteController
from bigeye_sdk.exceptions.exceptions import FileLoadException from bigeye_sdk.exceptions.exceptions import FileLoadException
from bigeye_sdk.model.big_config import BigConfig, TableDeployment, TableDeploymentSuite from bigeye_sdk.model.big_config import (
BigConfig,
ColumnSelector,
RowCreationTimes,
TableDeployment,
TableDeploymentSuite,
TagDeployment,
TagDeploymentSuite,
)
from bigeye_sdk.model.protobuf_message_facade import ( from bigeye_sdk.model.protobuf_message_facade import (
SimpleCollection, SimpleCollection,
SimpleMetricDefinition, SimpleMetricDefinition,
@ -130,6 +138,122 @@ def deploy(
print("No metadata file for: {}.{}.{}".format(project, dataset, table)) print("No metadata file for: {}.{}.{}".format(project, dataset, table))
def _update_table_bigconfig(
bigconfig,
metadata,
project,
dataset,
table,
):
"""Update the BigConfig file to monitor a table."""
default_metrics = [
SimplePredefinedMetricName.FRESHNESS,
SimplePredefinedMetricName.VOLUME,
]
for collection in bigconfig.table_deployments:
for deployment in collection.deployments:
for metric in deployment.table_metrics:
if metric.metric_type.predefined_metric in default_metrics:
default_metrics.remove(metric.metric_type.predefined_metric)
if metadata.monitoring.collection and collection.collection is None:
collection.collection = SimpleCollection(
name=metadata.monitoring.collection
)
if len(default_metrics) > 0:
deployments = [
TableDeployment(
fq_table_name=f"{project}.{project}.{dataset}.{table}",
table_metrics=[
SimpleMetricDefinition(
metric_type=SimplePredefinedMetric(
type="PREDEFINED", predefined_metric=metric
),
metric_schedule=SimpleMetricSchedule(
named_schedule=SimpleNamedSchedule(
name="Default Schedule - 13:00 UTC"
)
),
)
for metric in default_metrics
],
)
]
collection = None
if metadata.monitoring.collection:
collection = SimpleCollection(name=metadata.monitoring.collection)
bigconfig.table_deployments += [
TableDeploymentSuite(deployments=deployments, collection=collection)
]
def _update_view_bigconfig(
bigconfig,
metadata,
project,
dataset,
table,
):
"""Update the BigConfig file to monitor a view."""
default_metrics = [
SimplePredefinedMetricName.FRESHNESS_DATA,
SimplePredefinedMetricName.VOLUME_DATA,
]
for collection in bigconfig.tag_deployments:
for deployment in collection.deployments:
for metric in deployment.metrics:
if metric.metric_type.predefined_metric in default_metrics:
default_metrics.remove(metric.metric_type.predefined_metric)
if metadata.monitoring.collection and collection.collection is None:
collection.collection = SimpleCollection(
name=metadata.monitoring.collection
)
if len(default_metrics) > 0:
deployments = [
TagDeployment(
column_selectors=[
ColumnSelector(name=f"{project}.{project}.{dataset}.{table}.*")
],
metrics=[
SimpleMetricDefinition(
metric_type=SimplePredefinedMetric(
type="PREDEFINED", predefined_metric=metric
),
metric_schedule=SimpleMetricSchedule(
named_schedule=SimpleNamedSchedule(
name="Default Schedule - 13:00 UTC"
)
),
)
for metric in default_metrics
],
)
]
collection = None
if metadata.monitoring.collection:
collection = SimpleCollection(name=metadata.monitoring.collection)
bigconfig.tag_deployments += [
TagDeploymentSuite(deployments=deployments, collection=collection)
]
bigconfig.row_creation_times = RowCreationTimes(
column_selectors=[
ColumnSelector(
name=f"{project}.{project}.{dataset}.{table}.{metadata.monitoring.partition_column}"
)
]
)
@monitoring.command( @monitoring.command(
help=""" help="""
Update BigConfig files based on monitoring metadata. Update BigConfig files based on monitoring metadata.
@ -158,55 +282,22 @@ def update(name: str, sql_dir: Optional[str], project_id: Optional[str]) -> None
else: else:
bigconfig = BigConfig(type="BIGCONFIG_FILE") bigconfig = BigConfig(type="BIGCONFIG_FILE")
default_metrics = [ if (metadata_file.parent / VIEW_FILE).exists():
SimplePredefinedMetricName.FRESHNESS, _update_view_bigconfig(
SimplePredefinedMetricName.VOLUME, bigconfig=bigconfig,
] metadata=metadata,
project=project,
for collection in bigconfig.table_deployments: dataset=dataset,
for deployment in collection.deployments: table=table,
for metric in deployment.table_metrics: )
if metric.metric_type.predefined_metric in default_metrics: else:
default_metrics.remove( _update_table_bigconfig(
metric.metric_type.predefined_metric bigconfig=bigconfig,
) metadata=metadata,
project=project,
if metadata.monitoring.collection and collection.collection is None: dataset=dataset,
collection.collection = SimpleCollection( table=table,
name=metadata.monitoring.collection )
)
if len(default_metrics) > 0:
deployments = [
TableDeployment(
fq_table_name=f"{project}.{project}.{dataset}.{table}",
table_metrics=[
SimpleMetricDefinition(
metric_type=SimplePredefinedMetric(
type="PREDEFINED", predefined_metric=metric
),
metric_schedule=SimpleMetricSchedule(
named_schedule=SimpleNamedSchedule(
name="Default Schedule - 17:00 UTC"
)
),
)
for metric in default_metrics
],
)
]
collection = None
if metadata.monitoring.collection:
collection = SimpleCollection(
name=metadata.monitoring.collection
)
bigconfig.table_deployments += [
TableDeploymentSuite(
deployments=deployments, collection=collection
)
]
bigconfig.save( bigconfig.save(
output_path=bigconfig_file.parent, output_path=bigconfig_file.parent,

Просмотреть файл

@ -8,5 +8,5 @@ description: |-
owners: owners:
- ascholtz@mozilla.com - ascholtz@mozilla.com
monitoring: monitoring:
enabled: false enabled: true
partition_column: date partition_column: date