Add mutations such as mutiplication/division to allow exposing metrics in the proper units

[frittentheke]

Is your feature request related to a problem? Please describe.
Some logs contain valuable metric data Promtail is able aggregate and expose as Prometheus metrics already.
Unfortunately the data is not always in the proper format / unit to follow the best-practices and to use the Prometheus recommended base-units (https://prometheus.io/docs/practices/naming/#base-units).
So a field extracted from a log could contain milliseconds instead of seconds or bits insteads of bytes.

Describe the solution you'd like
It would be great to have a few mutations such as multiplication or / division available which can be applied to fields.
Potentially this would could be covered by the template (https://grafana.com/docs/loki/latest/clients/promtail/stages/template/#supported-functions) feature and go-template. But unfortunately there are not arithmetic functions allowed / implemented (yet):

invalid template stage config: template: pipeline_template:1: function \"div\" not defined"

To not have to maintain a homegrown list of functions the Sprig library (http://masterminds.github.io/sprig/) could be added enabling more than enough magic to mutate things as required - i.e. use their divf on ms to become seconds.

Describe alternatives you've considered
This functionality cannot be covered by string manipulations such as the already available regex or the existing go template magic.

Additional context

[cyriltovena]

There's another issue somewhere for this. I prefer to add slowly what we need instead of using sprig, this is mostly because it is run on server side.

[frittentheke]

@cyriltovena thanks for your response !

I suppose you are talking about:

• Adding Math functions to line_format #2847
• Allow math on label_format expressions #2955

which are technically related, but they are more about LogQL than the metrics extraction.
In any case I suppose once implemented, those functions are likely applicable to all those cases.

[stale]

This issue has been automatically marked as stale because it has not had any activity in the past 30 days. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.

[frittentheke]

not stale

[frittentheke]

@cyriltovena I believe this feature was implemented via merged PR #3434 ?

[cyriltovena]

only for LogQL so far.