wilfred-s commented on PR #579: URL: https://github.com/apache/yunikorn-core/pull/579#issuecomment-1629976789
If we can do a reload without a huge impact we should try and make it configurable. Looking at the base code we have I think we can. The code has already got a nice way to copy all data and set it aside in a new slice. That code can be reused to copy all the data and unwrap it before using it as the base for the new ring buffer content. It even allows for the disposal of records that are too old to keep in the shrink case. A single copy after that and setting everything beside the `id` correctly would finish up the work. Based on this we might need to run some performance checks. See how long it takes to grow or shrink a full large 1, 2 and 5 million entries buffer. See how big the overhead is in memory usage and how long it takes, The overhead should be limited as we only have pointers in the slice but a check is needed. -- 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]
