query-exporter - Export Prometheus metrics from SQL queries

query-exporter is a Prometheus exporter which allows collecting metrics from database queries, at specified time intervals.

It uses SQLAlchemy to connect to different database engines, including PostgreSQL, MySQL, Oracle and Microsoft SQL Server.

Each query can be run on multiple databases, and update multiple metrics.

The application is called with a configuration file that looks like this:

databases:
  db1:
    dsn: sqlite://
  db2:
    dsn: sqlite://

metrics:
  metric1:
    type: gauge
    description: A sample gauge
  metric2:
    type: summary
    description: A sample summary
  metric3:
    type: histogram
    description: A sample histogram
    buckets: [10, 20, 50, 100, 1000]
  metric4:
    type: enum
    description: A sample enum
    states: [foo, bar, baz]

queries:
  query1:
    interval: 5
    databases: [db1]
    metrics: [metric1]
    sql: SELECT random() / 1000000000000000
  query2:
    interval: 20
    databases: [db1, db2]
    metrics: [metric2, metric3]
    sql: |
      SELECT abs(random() / 1000000000000000),
             abs(random() / 10000000000000000)
  query3:
    interval: 10
    databases: [db2]
    metrics: [metric4]
    sql: |
      SELECT value FROM (
        SELECT 'foo' AS value UNION
        SELECT 'bar'
        UNION SELECT 'baz')
      ORDER BY random()
      LIMIT 1

The dsn connection string has the following format:

dialect[+driver]://[username:password][@host:port]/database

(see SQLAlchemy documentation for details on the available options).

The metrics list in the query configuration must match values returned by the query defined in sql.

The interval value is interpreted as seconds if no suffix is specified; valid suffix are s, m, h, d. Only integer values can be specified. If no value is specified (or specified as null), the query is executed at every HTTP request.

Queries will usually return a single row, but multiple rows are supported, and each row will cause an update of the related metrics. This is relevant for any kind of metric except gauges, which will be effectively updated to the value from the last row.

For the configuration above, exported metrics look like this:

# HELP metric1 A sample gauge
# TYPE metric1 gauge
metric1{database="db1"} 1549.0
# HELP metric2 A sample summary
# TYPE metric2 summary
metric2_count{database="db2"} 1.0
metric2_sum{database="db2"} 5229.0
metric2_count{database="db1"} 1.0
metric2_sum{database="db1"} 4513.0
# TYPE metric2_created gauge
metric2_created{database="db2"} 1.5456472955657206e+09
metric2_created{database="db1"} 1.5456472955663064e+09
# HELP metric3 A sample histogram
# TYPE metric3 histogram
metric3_bucket{database="db2",le="10.0"} 0.0
metric3_bucket{database="db2",le="20.0"} 0.0
metric3_bucket{database="db2",le="50.0"} 0.0
metric3_bucket{database="db2",le="100.0"} 0.0
metric3_bucket{database="db2",le="1000.0"} 1.0
metric3_bucket{database="db2",le="+Inf"} 1.0
metric3_count{database="db2"} 1.0
metric3_sum{database="db2"} 714.0
metric3_bucket{database="db1",le="10.0"} 0.0
metric3_bucket{database="db1",le="20.0"} 0.0
metric3_bucket{database="db1",le="50.0"} 0.0
metric3_bucket{database="db1",le="100.0"} 0.0
metric3_bucket{database="db1",le="1000.0"} 1.0
metric3_bucket{database="db1",le="+Inf"} 1.0
metric3_count{database="db1"} 1.0
metric3_sum{database="db1"} 602.0
# TYPE metric3_created gauge
metric3_created{database="db2"} 1.545647295565831e+09
metric3_created{database="db1"} 1.5456472955663848e+09
# HELP metric4 A sample enum
# TYPE metric4 gauge
metric4{database="db2",metric4="foo"} 0.0
metric4{database="db2",metric4="bar"} 1.0
metric4{database="db2",metric4="baz"} 0.0

Metrics are automatically tagged with the database label so that indipendent series are generated for each database.

Database engines

SQLAlchemy doesn't depend on specific Python database modules at installation. This means additional modules might need to be installed for engines in use, as follows:

pip install SQLAlchemy[postgresql] SQLAlchemy[mysql] ...

based on which databased is in use.

See supported databases for details.

Carto extension

You can define a carto connection instead of a SQL DSN. If you want to do so, use a carto: entry in your database.

Example:

databases:
  test_carto:
    carto:
      user: my_carto_user
      api_key: my_carto_api_key

metrics:
  observations_simple_count:
    type: gauge
    description: Simple count to check if this works...

queries:
  query_count_simple_count:
    interval: 120s
    databases: [test_carto]
    metrics: [observations_simple_count]
    sql: SELECT count(*) from county_population;

• You cannot use both dsn and carto entries in the same database as that makes no sense.
• The available fields for the configuration object are the same as for the Longitude CartoDataSource objects.
• As of today, such fields are (keep in mind that some might not make sense for monitoring):
  • api_version: v2 by default
  • uses_batch: False by default
  • on_premise_domain: '' by default. If provided, the Carto URL will use it. If not, the default user URL will.
  • api_key: '' by default. Mandatory. Master api key recommended.
  • user: '' by default. Mandatory. CARTO user (not email)
  • cache: Empty by default. Cache configuration. Useless in this context for now.

Development environment

The easiest way to install the required dependencies is to create a virtual environment and install the package:

python setup.py install pipenv install -e .