How to rename label within a metric in Prometheus

In Prometheus, renaming a label typically involves using the label_replace() function in PromQL (Prometheus Query Language) queries. This function performs regular expression replacement on the labels of the query results to achieve renaming of the labels.

Function Definition

label_replace() function's basic format is as follows:

plaintext
label_replace(v instant-vector, dst_label string, replacement string, src_label string, regex string)

• v: Input vector
• dst_label: Target label name, i.e., the new label name
• src_label: Source label name, i.e., the current label to be replaced
• regex: Regular expression used to match the source label values
• replacement: Replacement content, where you can specify the new label value

Example

Assume we have a metric http_requests_total with a label path, and we want to rename this label to endpoint.

The query is as follows:

promql
label_replace(http_requests_total, "endpoint", "$1", "path", "(.*)")

Here, "(.*)" is a regular expression that matches all possible values of the path label, and $1 assigns this value to the new endpoint label.

Application Scenarios

In practical applications, we may encounter scenarios where we need to standardize label names across different data sources. For example, suppose we collect data from two different systems, one using path as the label and the other using endpoint as the label. By using label_replace(), we can unify these labels into a single name, making it easier to integrate and compare data during queries and visualization.

Important Notes

• When using label_replace(), the regular expression must correctly match the values of the source label; otherwise, no replacement occurs.
• The replacement operation adds a new label to the query results rather than directly modifying the original data. The original data's label names and values remain unchanged.
• If performance issues arise during usage, consider whether label standardization and normalization should be performed during data collection or configuration phase to reduce the burden on queries.

By reasonably utilizing label_replace(), you can effectively manage and adjust labels in Prometheus, making data analysis and monitoring more flexible and accurate.