Pushgateway

Important

To save data to Pushgateway install garf-io with Pushgateway support

pip install garf-io[pushgateway]

pushgateway writer allows you to publish GarfReport to a Pushgateway endpoint.

clipython

garf query.sql --source API_SOURCE \
  --output pushgateway

from garf.core import report
from garf.io.writers import pushgateway_writer

sample_report = report.GarfReport(results=[[1]], column_names=['one'])

writer = pushgateway_writer.PushgatewayWriter()
writer.write(sample_report, 'grouping_key')

Parameters

endpoint

By default writer pushes data to http://localhost:9091/. You can overwrite it with endpoint parameter.

clipython

garf query.sql --source API_SOURCE \
  --output pushgateway \
  --pushgateway.endpoint=http://pushgateway:9091

from garf.core import report
from garf.io.writers import pushgateway_writer

sample_report = report.GarfReport(results=[[1]], column_names=['one'])

writer = pushgateway_writer.PushgatewayWriter(endpoint="http://pushgateway:9091")
writer.write(sample_report, 'grouping_key')

namespace

Every metric pushes to Pushgateway is prefix with garf_ prefix. You can overwrite it with namespace parameter.

clipython

garf query.sql --source API_SOURCE \
  --output pushgateway \
  --pushgateway.namespace=my_garf

from garf.core import report
from garf.io.writers import pushgateway_writer

sample_report = report.GarfReport(results=[[1]], column_names=['one'])

writer = pushgateway_writer.PushgatewayWriter(namespace="my_garf")
writer.write(sample_report, 'grouping_key')

job

By default all metrics is pushed into garf job. You can overwrite it with job parameter.

clipython

garf query.sql --source API_SOURCE \
  --output pushgateway \
  --pushgateway.job=garf_job

from garf.core import report
from garf.io.writers import pushgateway_writer

sample_report = report.GarfReport(results=[[1]], column_names=['one'])

writer = pushgateway_writer.PushgatewayWriter(job="garf_job")
writer.write(sample_report, 'grouping_key')

expose_metrics_with_zero_values

By default only metrics with non-zero values are pushed to Pushgateway. You can overwrite it with expose_metrics_with_zero_values parameter.

clipython

garf query.sql --source API_SOURCE \
  --output pushgateway \
  --pushgateway.expose_metrics_with_zero_values

from garf.core import report
from garf.io.writers import pushgateway_writer

sample_report = report.GarfReport(results=[[1]], column_names=['one'])

writer = pushgateway_writer.PushgatewayWriter(expose_metrics_with_zero_values=True)
writer.write(sample_report, 'grouping_key')

Query syntax

In order to expose metrics to Pushgateway you need to ensure that either field name or alias should contain metric in it. So given the query

SELECT
  dimension.name AS name,
  metric.name AS field1,
  name AS metric_field2
FROM resource

two metrics will be pushed to Pushgateway - <namespace>_<job>_field1 and <namespace>_<job>_field2.

Info metrics

If you query does not contains any metric all it's dimensions will be bundled into _info metric (i.e. <namespace>_<job>_info) with a value of 1.0. info metrics are useful when you need to capture the mapping between various attributes in an API but don't care about the actual value.