Skip to content

Custom Metrics

You can send custom metrics to any of your monitoring Apps. These metrics will be counted towards your daily data points quota which may trigger rejection in case of a huge amount of data points. The limits are set very high, though, and under the normal operation it is hard to reach them.

Metrics are sent to metrics receivers (in the US region to, in the EU region to Expected format is based on Influx Line Protocol, with a few specific additional requirements. That means that many existing libraries/tools that know how to send metrics to InfluxDB can also send the metrics to Sematext backend.

Sematext specifics

Sematext backend supports v1 of the Influx Line Protocol, so endpoint that accepts the metrics is Each metrics line sent to Sematext backend must contain a tag with the name token and value equal to the token of Sematext Monitoring App to which you want to send the metrics.

Additionally, it is strongly encouraged, where applicable, to add tag with value equal to the name of the monitored host. This tag is deeply integrated into Sematext backend and lack of it may cause some features to not work.

Sematext backend uses a catalogue of common tags. Depending on the nature of your custom metrics, adding some of the tags from this catalogue can make your metrics better integrated into Sematext ecosystem.

Metrics line example

appStats,token=00000000-1111-2222-3333-444444444444,,service=registration user.requests.count=10i,user.requests.time=327i 1628605794318000000

In this example we can see:

  • appStats - a measurement or a namespace, it describes the type of metrics in this particular line and it can be anything that makes sense to you. It will be used as a prefix to all metrics in this line (so, full name of a metric user.requests.count is actually appStats.user.requests.count)
  • token - a token of your Sematext Monitoring App
  • and service - two tags used to describe these metrics. You can append more tags by appending them after service=registration (tags have to be separated by a comma)
  • user.requests.count and user.requests.time - two metrics. You can append more metrics. Metrics need to be comma separated.
  • 1628605794318000000 - the third and the final element of a metric line is the measurement timestamp expressed in nanoseconds

A single request can contain any number of metric lines, each line added in a new line (\n as a separator between the lines). We recommend sending batches of between 100 and 500 lines of metrics. Of course, if you don't gather as many lines, but would still like the metrics to appear in dashboards as soon as possible, send smaller batches.

Example of sending a batch of metrics:

curl -XPOST -H "Content-Type: text/plain" "" -d '
appStats,token=00000000-1111-2222-3333-444444444444,,service=registration user.requests.count=10i,user.requests.time=327i 1628605794318000000
appStats,token=00000000-1111-2222-3333-444444444444,,service=payment user.requests.count=31i,user.requests.time=1152i 1628605797988000000
appStats,token=00000000-1111-2222-3333-444444444444,,service=email user.requests.count=20i,user.requests.time=218i 1628605812295000000


Each metric has attributes that describe how it is used in the UI. Sematext specific bit is that such info about metrics (we call it the "metainfo") can be sent to a separate endpoint. It is enough to send such metainfo just once for your application.

Metric attributes are:

  • name - in the example above, names are user.requests.count and user.requests.time
  • label - name that will be used in charts
  • description - longer description of the meaning of a particular metric
  • numericType - long or double
  • type - counter or gauge. In the example above, both metrics are counters because they describe how many requests there were and how much time those requests took since the previous measurement. Metrics like memory.used or which represent exact value of some metric at particular point in time are gauges.
  • unit - can be one of standard units (ms, sec, bytes...) or something custom that makes sense to you.

Here is an example of metainfo line:

appStats,label=requests\ count,numericType=long,token=00000000-1111-2222-3333-444444444444,type=counter user.requests.count

Metainfo entries can be sent in a single batch, each entry in a new line. The endpoint where these entries are sent is (and a similar one for EU region, where base server is

If the metainfo for some metric is missing, the following values will be used:

  • name - whatever is the name of the metric, e.g. user.requests.count
  • label - name will be used instead
  • description - it will be empty
  • numericType - double
  • type - gauge
  • unit - it will be empty

One of these attributes can be adjusted directly in the charts - type. There is a rollup function which is a SUM in case of a counter and an AVG in case of gauge metric.

Libraries and tools

A list of official libraries to connect to InfluxDB is here. These are, however, libraries that connect to 1.8/2.0 version of InfluxDB, while Sematext backend requires 1.7 or older. Github repositories of these libraries provide a link to older version which supports 1.7 and which would work with Sematext.

Another possibility is to use a tool like Telegraf Agent. It has support for InfluxDB 1.7, but (at the time of this writing) there is also a pull-request for fully compatible Sematext Output. Either of these can be used to send the metrics to Sematext backend, but using Sematext Output is much simpler as you wouldn't have to worry about any Sematext specific, everything is encapsulated in Sematext Output logic.