nickva opened a new pull request, #6026: URL: https://github.com/apache/couchdb/pull/6026
Previously, even when user picked a non-default compression algorithm in some instances, like the purge trees and emsort, snappy would still be used such. Make sure to always respect user's choice, and if they picked something non-default always use that. [1] There is a good reason to avoid snappy as doesn't properly yield when encoding or decoding. In a low concurrency environment it might not matter, but as the concurrency ramps up it could lead to scheduler collapse (it could manifest as high p90+ latencies while cpus usage remains low). A similar effect is illustrated in scheduler responsiveness test for jiffy vs other C json libraries in https://github.com/nickva/bench#scheduler-responsiveness-test -- 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]
