Docker tag latest
is based on the same commit as the latest released version.
Service that generates basic metrics for SQL result sets and exposing them as Prometheus metrics.
This service relies on the SQL Agent service to execute and return the SQL result sets.
- Static configuration files are used to define the queries to monitor.
- Each query has a designated worker for execution.
- An interval is used to define how often to execute the query.
- Failed queries are automatically retried using a backoff mechanism.
- Faceted metrics are supported.
- A single metric's different facets can be filled in from different data sources.
- Metric names are exposed in the format
query_result_<metric name>
. - With faceted metrics, the name of the data column is determined by the
data-field
key in config, and all other columns (and column values) are exposed as labels. - If the result set consists of a single row and column, the metric value is obvious and
data-field
is not needed. - Label names under the same metric should be consistent.
- Each different query (query entry in config) for the same metric should lead to different label values.
Usage of prometheus-sql:
-config string
Configuration file to define common data sources etc.
-host string
Host of the service. (0.0.0.0)
-lax
Tolerate invalid files in queryDir
-port int
Port of the service. (default 8080)
-queries string
Path to file containing queries. (default "queries.yml")
-queryDir string
Path to directory containing queries.
-service string
Query of SQL agent service.
A queries file is required for the application to know which data source to query and which queries that shall be monitored.
In the repository there is an example file that you can have a look at.
The config file is optional and can defined some default values for queries and data sources which can be referenced by queries. The benefit of referencing a data source will be reduction of duplication of database connection information. See example config file here and queries file which utilizes the config information.
Create a queries.yml
file in the current directory and run the following:
prometheus-sql
or for an alternate path, use the -queries or the -queryDir option:
prometheus-sql -queries ${PWD}/queries.yml
Run the SQL agent service.
docker run -d --name sqlagent dbhi/sql-agent
Run this service. Mount the queries.yml
file and link the SQL Agent service.
docker run -d \
--name prometheus-sql \
-p 8080:8080 \
-v ${PWD}/queries.yml:/queries.yml \
--link sqlagent:sqlagent \
dbhi/prometheus-sql
If you want to separate database connection information etc you can do that by specifying data sources in separate file which you then can mount:
docker run -d \
--name prometheus-sql \
-p 8080:8080 \
-v ${PWD}/queries.yml:/queries.yml \
-v ${PWD}/prometheus-sql.yml:/prometheus-sql.yml \
--link sqlagent:sqlagent \
dbhi/prometheus-sql \
-service http://sqlagent:5000 \
-config prometheus-sql.yml
To view a plain text version of the metrics, open up the browser to the http://localhost:8080/metrics (or http://192.168.59.103:8080/metrics for boot2docker users).
Alternately, use the docker-compose.yml
file included in this repository. The volumes
section be added for mounting the queries.yml
file.
Read instructions how to contribute before you start.
How do you I provide additional options to the database connection?
Additional options are set in the config.yml
file, specifically as additional key-value pairs in the connection map. These are passed to the SQL Agent service which construct a DSN string to establish the connection on the backend (alternately you can set the dsn
key as the full string).
As an example, a common gotcha when using Postgres in a development environment is to ignore SSL not being enabled on the server. This can be done by adding the sslmode: disable
option in the connection map.
# ...
connection:
host: example.org
port: 5432
user: postgres
password: s3cre7
database: products
sslmode: disable