josh-mckenzie commented on PR #1954: URL: https://github.com/apache/cassandra/pull/1954#issuecomment-1331256000
Some top level questions (about to transfer line items from notes to the diff): * Why would caching a mutation at any given size be particularly bad for GC? We're already paying the tax to ser or deser in the first place; how long do we expect these messages to hang around and live -> worst-case is timeout from replicas? * Where does our initial value for the default on `cacheable_mutation_size_limit_bytes` come from? * If we're doing ThreadLocal on the limitedScratchBuffer, how are we managing the implications of a Mutation `deserialize` call bouncing across threads? We'll end up with a deserialized copy of it living in memory on another thread should something jump to another Stage in the future etc right? Not a big deal as they should be short-lived eden and we'd be deserializing it anyway, just thinking of the tradeoffs between having a central cache of deserialized mutations w/the current "raciness" attributes vs. the threadlocal approach. -- 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]
