influxdb's data schema supports fields and tags. tags are indexed and fields are not.
prometheus support labels and float64 value.
Mapping between prometheus and influxdb data is as follows:
labels : tags
float64 : a default field
My use case is monitoring batch jobs having auxillary metadata (processid,logfilepath etc.). I dont want these attributes to be indexed.
How do I model these attributes in prometheus ? If I use labels for the same, its going to explode memory usage (label indexing), since its values are not enumerated.
Whats the best way to take care above attributes in prometheus ?
As the processid is a number, you can expose it as a gauge. For strings you can expose them as a label on a 1-valued gauge, such as is done for software versions.