resize the memory pool. This only adjusts the Pool size. It does not release the slabs back to the SlabAllocator if the new size is less than the curr

* Exports the current state as a thrift object for later restoration. */

Shrink buffer logical size (doesn't reallocate)

Return a new view from part of this current view. This view must NOT be null. @param offset start of the data to encapsulate @param size size of b

* Estimate percentiles for reporting * @return PercentileStats::Estimates with nanosecond latency percentiles */

Convert to semi future.

Moves ownership into a shared_ptr. This is irreversible. TODO: in the future, we should consider if we want to allow user to convert shared_ptr back i

This function updates the current itme stamp. It blocks the thread if the timeStampSecond belongs to a future bucket and some other threads haven't fi

Return true if this handle couldn't be fulfilled because the item had already expired. If item is present then the source of truth lies with the actua

* Acquire a write (exclusive) lock on the mutex associated with a key. The * returned WriteLock automatically releases the lock when destroyed.

* Add the passed-in task to the FiberManager. * * @param func Task functor; must have a signature of `void func()`. * The object

Return a number that's equal or smaller than @num and aligned on @alignment

Return a number that's equal or bigger than @num and aligned on @alignment

If true it means there's no valid cache or parent handle assocaited with this BufferManager, no operation is valid.

device settings

Return the readonly memory

return the usable size in bytes for this allocator.

Expose the admission policy for testing.

return the usable size of the pool. Usable pool size is configured pool size minus advised away size

get a snapshot across all instances.

Get the current total object size limit used by the size controller.

Calculates hit ratio, based on total operations and total misses @param ops total operations @param miss operations that missed

Initialise the LRU cache based on the supplied config. This should be called once and once only per execution

marks the token as invalidated. This will ensure that we dont execute any function on this token and simply remove the token when the token gets destr

returns the current state of whether nvmcache is enabled or not. nvmcache can be disabled if the backend implementation ends up in a corrupt state

returns the advised memory in the unit of slabs.

Get the number of valid entries