Skip to content


gnmic supports exporting subscription updates to influxDB time series database


An influxdb output can be defined using the below format in gnmic config file under outputs section:

    # required
    type: influxdb 
    # influxDB server address
    url: http://localhost:8086 
    # empty if using influxdb1.8.x
    org: myOrg 
    # string in the form database/retention-policy. Skip retention policy for the default on
    bucket: telemetry
    # influxdb 1.8.x use a string in the form: "username:password"
    # number of points to buffer before writing to the server
    batch-size: 1000 
    # flush period after which the buffer is written to the server whether the batch_size is reached or not
    flush-timer: 10s
    # if true, the influxdb client will use gzip compression in write requests.
    use-gzip: false
    # (deprecated, use tls.skip-verify: true) 
    #if true, the influxdb client will use a secure connection to the server.
    enable-tls: false
    # tls config
      # string, path to the CA certificate file,
      # this will be used to verify the clients certificates when `skip-verify` is false
      # string, client certificate file.
      # string, client key file.
      # boolean, if true, the client will not verify the server
      # certificate against the available certificate chain.
      skip-verify: false
    # boolean, if true the message timestamp is changed to current time
    override-timestamps: false 
    # server health check period, used to recover from server connectivity failure.
    # health check is disabled by default, can be enabled by setting the below field to any value other that zero.
    # with a minimum allowed period of 30s.
    health-check-period: 0s 
    # defines the write timestamp precision, 
    # one of `s` for second, `ms` for millisecond, `us` for microsecond and `ns` for nanoseconds
    # any other value defaults to `ns`.
    timestamp-precision: ns
    # server health check period, used to recover from server connectivity failure
    health-check-period: 30s 
    # enable debug
    debug: false 
    # string, one of `overwrite`, `if-not-present`, ``
    # This field allows populating/changing the value of Prefix.Target in the received message.
    # if set to ``, nothing changes 
    # if set to `overwrite`, the target value is overwritten using the template configured under `target-template`
    # if set to `if-not-present`, the target value is populated only if it is empty, still using the `target-template`
    # string, a GoTemplate that allow for the customization of the target field in Prefix.Target.
    # it applies only if the previous field `add-target` is not empty.
    # if left empty, it defaults to:
    # {{- if index . "subscription-target" -}}
    # {{ index . "subscription-target" }}
    # {{- else -}}
    # {{ index . "source" | host }}
    # {{- end -}}`
    # which will set the target to the value configured under `subscription.$` if any,
    # otherwise it will set it to the target name stripped of the port number (if present)
    # NOT IMPLEMENTED boolean, enables the collection and export (via prometheus) of output specific metrics
    enable-metrics: false 
    # list of processors to apply on the message before writing
    event-processors: []
    # cache, if present enables the influxdb output to cache received updates and write them all together 
    # at `cache-flush-timer` expiry.
      # duration, if > 0, enables the expiry of values written to the cache.
      expiration: 0s
      # debug, if true enable extra logging
      debug: false
    # cache-flush-timer
    cache-flush-timer: 5s

gnmic uses the event format to generate the measurements written to InfluxDB. When an event has been processed through gnmic processors, the final value of the subscription-name tag will be used as an InfluxDB measurement name and the tag will be removed. If the subscription-name tag does not exist in the event, the event's Name will be used as InfluxDB measurement.


When caching is enabled, the received messages are not written directly to InfluxDB, they are first cached as gNMI updates and written in batch when the cache-flush-timer is reached.

The below diagram shows how an InfluxDB output works with and without cache enabled:

When caching is enabled, the cached gNMI updates are periodically retrieved in batch, converted to events.

If processors are defined under the output, they are applied to the whole list of events at once. This allows augmenting some messages with values from other messages even if they where collected from a different target/subscription.