@dhruvilshah3 Thanks for the comments. I get your ideas but I wonder whether it actually simplifies this patch. Currently, the memory map operation and the initialization of all the fields of a index object are done when the index object is created. Even though we can maintain a flag in `LogSegment` layer, we still need to add similar logic in the index classes to allow on-demand memory map and initialization. What do you think?
I think the logic in this PR is understandable and not very complicated. Also, this patch is already been tested in our production environment for almost two months and no issues have been seen so we have some confidence in terms of correctness and performance for this PR. [ Full content available at: https://github.com/apache/kafka/pull/5498 ] This message was relayed via gitbox.apache.org for [email protected]
