prometheus-influxdb-adapter

README

prometheus-influxdb-adapter

Prometheus remote storage adapter for InfluxDB.

Usage

./prometheus-influxdb-adapter

Read

Upon receiving a request from Prometheus, prometheus-influxdb-adapter sends the requet for reading samples to the InfluxDB RPC server. The address of the RPC server (default is :8082) can be specified using --read.influxdb.rpc-address flag.

./prometheus-influxdb-adapter --read.influxdb.rpc-address=influxdb.example.com:8082

Points are read from prometheus database and the default retention policy by default.

./prometheus-influxdb-adapter --read.influxdb.database db1 --read.influxdb.retention-policy=policy1

If you specify multiple retention policies prefixed with duration, prometheus-influxdb-adapter selects one based on the duration and the interval step of the request. For example,

./prometheus-influxdb-adapter --read.influxdb.retention-policy=1h:1h --read.influxdb.retention-policy=5m:5m --read.influxdb.retention-policy=1m

1h retention policy is used if the interval step of the request is smaller than 1 hour, and 5m retention policy is used if the step is bigger than 5 minutes and smaller than 1 hour and 1m is used if the step is smaller than 1 minute or not given.

Note: The interval step in the request is available only if you are running the Prometheus server built from master at the moment.

You can also change the field (default is value field) in InfluxDB to read the value from.

./prometheus-influxdb-adapter --read.influxdb.field=field1

If you prefix the field with a Prometheus's function name, the field is used only if the function name given in the request matches the function name specified.

./prometheus-influxdb-adapter --read.influxdb.field=max_over_time:max --read.influxdb.field=min_over_time:min --read.influxdb.field=sum_over_time:sum --read.influxdb.field=mean

Note: The function name in the request is available only if you are running the Prometheus server built from master at the moment.

Write

By default, prometheus-influxdb-adapter sends points to http://localhost:8086. You can change it using the --write.influxdb.url flag.

./prometheus-influxdb-adapter --write.influxdb.url=http://influxdb.example.com:8086

Points are written into prometheus database with the default retention policy by default.

./prometheus-influxdb-adapter --write.influxdb.database=db1 --wirte.influxdb.retention-policy=policy1

You can also change the field (default is value field) in InfluxDB for storing the value.

./prometheus-influxdb-adapter --write.influxdb.field=field1

Flags

$ ./prometheus-influxdb-adapter --help
Usage of ./prometheus-influxdb-adapter:
  -log.level string
        Only log messages with the given severity or above. One of: [debug, info, warn, error] (default "info")
  -read.influxdb.database string
        InfluxDB database to read points from. (default "prometheus")
  -read.influxdb.field value
        Field to read sample values from. (default value)
  -read.influxdb.retention-policy value
        Retention policy to query from.
  -read.influxdb.rpc-address string
        Address of InfluxDB RPC server to receive points from. (default ":8082")
  -version
        Print version information.
  -web.listen-address string
        Address to listen on for web interface and telemetry. (default ":9201")
  -web.telemetry-path string
        Path under which to expose metrics. (default "/metrics")
  -write.influxdb.consistency-level string
        Default consistency level of the write operation.
  -write.influxdb.database string
        InfluxDB database to write points into. (default "prometheus")
  -write.influxdb.field string
        Field to store sample values in InfluxDB. (default "value")
  -write.influxdb.retention-policy string
        Default retention policy to use for points written to InfluxDB.
  -write.influxdb.timeout duration
        Timeout for sending points to InfluxDB. (default 30s)
  -write.influxdb.url string
        URL of InfluxDB server to send points to. (default "http://localhost:8086")

Metrics

Name	Help	Labels
dropped_samples_total	Total number of dropped samples.
http_request_duration_seconds	Histogram of HTTP request latencies.	handler
http_request_size_bytes	Histogram of HTTP request size.	handler
http_requests_total	Total number of HTTP requests.	handler, code
http_response_size_bytes	Histogram of HTTP response size.	handler
read_points_total	Total number of points read from InfluxDB.	database, retention_policy
received_samples_total	Total number of received samples.
written_points_total	Total number of points written to InfluxDB.	database, retention_policy

Docker

You can deploy this adapter using the kaorimatz/prometheus-influxdb-adapter Docker image.

docker run -d -p 9201:9201 kaorimatz/prometheus-influxdb-adapter

Development

Building

make

Building Docker image

make docker