Partial label matching in selects

[weeco]

Unfortunately, it currently only returns data points with an exact match of the label combination, which means, let's say you run storage.Select("consumergroup_topic_lag", []tstorage.Label{"consumer_group": "foo"}) then:

• consumergroup_topic_lag {"consumer_group": "foo"} will be returned
• consumergroup_topic_lag {} will not be returned
• consumergroup_topic_lag {"consumer_group": "foo", "another_label": "bar"} will not be returned

I think it would be beneficial if we had the option to query for metrics using partial labels. Example:

storage.Select("consumergroup_topic_lag", []tstorage.Label{"consumer_group": "foo"})

Returns all metrics where a label with key consumer_group equals to foo , regardless what other labels are set as long as the queried matches exist in the target:

consumergroup_topic_lag {"consumer_group": "foo", "topic": "bla"} 1000
consumergroup_topic_lag {"consumer_group": "foo", "topic": "bar"} 2000
consumergroup_topic_lag {"consumer_group": "foo", "topic": "basket"} 0

[nakabonne]

Ref: #36 (comment)

[nakabonne]

Currently, in order to simplify the file format, it calculates a hash value from labels and a metric name, and data points are stored close to each other within the disk using it as a key.

What we need to do first is investigate how other TSDBs deal with label indexes. I'll look into it when I have time 👍