You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kudu.apache.org by "Marton Greber (Jira)" <ji...@apache.org> on 2025/08/19 08:52:00 UTC
[jira] [Updated] (KUDU-3690) Implement query parameter-based filtering for the /metrics_prometheus webserver end-point
[ https://issues.apache.org/jira/browse/KUDU-3690?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Marton Greber updated KUDU-3690:
--------------------------------
Parent: KUDU-3691
Issue Type: Sub-task (was: Improvement)
> Implement query parameter-based filtering for the /metrics_prometheus webserver end-point
> -----------------------------------------------------------------------------------------
>
> Key: KUDU-3690
> URL: https://issues.apache.org/jira/browse/KUDU-3690
> Project: Kudu
> Issue Type: Sub-task
> Components: master, metrics, tserver
> Reporter: Alexey Serbin
> Priority: Major
>
> [Prometheus|https://prometheus.io/] supports optional HTTP query parameters for its scrape URLs, as per the [scrape_config section docs|https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config].
> With that, it's possible to implement filtering of Kudu metrics while scraping the {{/metrics_prometheus}} HTTP endpoint of the embedded webserver with Prometheus. In this context, it makes sense to have parity with the [filtering|https://kudu.apache.org/docs/administration.html#_collecting_metrics_via_http] of the JSON-formatted Kudu metrics sourced at the {{/metrics}} endpoint. At least, filtering by the following criteria should be implemented:
> * severity level
> * entity type
> * entity ID
> * set of substrings to match metric name
> * key/value pairs to match attributes of a metric (e.g., that's to filter metrics for all tablets that comprise a particular table identified by either "table_id" or "table_name" attribute)
--
This message was sent by Atlassian Jira
(v8.20.10#820010)