zhtaoxiang commented on PR #10104: URL: https://github.com/apache/pinot/pull/10104#issuecomment-1382287559
> Thanks, that makes the goal more understandable. I am OK with moving things around, but ew need to make absolutely sure that metrics do not somehow end up changing their name, or disappearing, or start to emit different values, etc. LinkedIn runs production use cases with hundreds of site-facing tables and thousands of internal tables. We rely on these metrics to alert us to any faults. Nobody looks at these tables manually, and they continue to serve millions of QPS everyday. > > Missing metrics or changing metric names can lead to not alerting when need, or trigger mass alerts that can cause a lot of manual work that we have not planned for -- causing operational headaches. > > Let us limit changes that are gratis. > > cc: @sajjad-moradi , @siddharthteotia @mcvsubbu Thanks for providing the feedback. Your points make a lot of sense! We definitely do not want to break production system using current metrics! What I did (see my previous PRs) and will do is to deprecate old methods (but I will not change or remove publicly available methods, except for those marking as visible for testing) and add new ones. We know that the migration from old methods to new ones will not happen over night and will take time. Util new methods are verified in production and we are confident, we will not replace old ones with new ones. What do you think? -- 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]
