richardstartin opened a new pull request #8227: URL: https://github.com/apache/pinot/pull/8227
We frequently see that prometheus is a resource hog in our customer deployments, with regular expression matching leading to excessive allocation and CPU time. Pinot currently uses version 0.12.0, which is hardcoded into deployments and the jar is checked in. <img width="1610" alt="Screenshot 2022-02-17 at 14 31 09" src="https://user-images.githubusercontent.com/16439049/154712664-ff9917f3-dc4d-4294-a18e-18971a4a502c.png"> <img width="1602" alt="Screenshot 2022-02-18 at 08 43 34" src="https://user-images.githubusercontent.com/16439049/154712720-04708ab9-d422-44fa-a616-7bf7e8e5a252.png"> Prometheus fixed this here, allowing : https://github.com/prometheus/jmx_exporter/pull/518 This PR changes this so that the latest version (pinned at 0.16.1) of jmx_prometheus_javaagent is downloaded in the docker image, and caching is enabled for our rules. **NOTE: scripts which are coupled with a specific version of prometheus:** ``` export JAVA_OPTS="-javaagent:jmx_prometheus_javaagent-0.12.0.jar ``` **will no longer enable metrics collection, but this PR means they will never need to break again. The migration path is to set:** ``` export JAVA_OPTS="-javaagent:jmx_prometheus_javaagent.jar ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
