MapStore Coalescing Write-Behind Batching produces too small batches

[Lukas Blunschi]

Hi,

we are using coalescing write-behind MapStores and we noticed a non-optimal behavior when the write-behind processor schedules the data to be written. It produces very small batches when there is a mix of store and delete operations.

I believe there would be a small fix to considerably improve this behavior and I already implemeted a PR for it... but since 3 weeks nobody had a look at the PR.

What do you think about this PR?

https://github.com/hazelcast/hazelcast/pull/24672

I also described the problem in an issue here:

https://github.com/hazelcast/hazelcast/issues/24763

Thanks for checking and best,

Lukas