franz1981 commented on issue #2845: ARTEMIS-2336 Use zero copy to replicate journal/page/large message file (AGAIN) URL: https://github.com/apache/activemq-artemis/pull/2845#issuecomment-535786097 @wy96f thanks to have tried! It sounds strange to me: I was thinking the reason why was taking more was due to being continuosly stopped/being awaken and sending short chunks to the network (ie more syscalls with less data). I have a strong feeling that maybe the point is that sendFile with send a 1 MB chunk *directly* without using the TCP buffer at all....if is the case, it means we should increase the chunkSize too (1 MB or at least 100K) and the TCP buffer accordingly... Did you observe that the network was saturated in both cases? If you use async-profiler you can check what the kernel does and where most the cost is for both cases....
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
