dengfengjiang77 commented on PR #16839: URL: https://github.com/apache/lucene/pull/16839#issuecomment-6504276978
I did a prod simulation test to double check this logic, it shows great that we remove entires the time in 2nd read and add a bit of time for accumulated centroid during the raw copy: Run 3 result for one full partition : | Phase | Mainline second-pass read | In-copy accumulate | Saved | |---|---|---|---| | Backfill (~2.5 h) | 458 s | 57.9 s | 400 s | | Transition (next 1 h) | 106 s | 9.4 s | 97 s | | Steady state (~6.8 h) | 73 s | 26.9 s | 46 s | | Total (~10 h) | 637 s | 94 s | 543 s | -- 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]
