_rebalance_find_msgs(self, keys: 'Set[Hashable] | None', workers: 'Iterable[WorkerState]') -> 'list[tuple[WorkerState, WorkerState, TaskState]]'
This method only defines the work to be performed; it does not start any network transfers itself.
The big-O complexity is O(wt + ke*log(we)), where
wt is the total number of workers on the cluster (or the number of whitelisted workers, if explicitly stated by the user)
we is the number of workers that are eligible to be senders or recipients
kt is the total number of keys on the cluster (or on the whitelisted workers)
ke is the number of keys that need to be moved in order to achieve a balanced cluster
There is a degenerate edge case O(wt + kt*log(we)) when kt is much greater than the number of whitelisted keys, or when most keys are replicated or cannot be moved for some other reason.
Returns list of tuples to feed into _rebalance_move_data:
sender worker
recipient worker
task to be transferred
Identify workers that need to lose keys and those that can receive them, together with how many bytes each needs to lose/receive. Then, pair a sender worker with a recipient worker for each key, until the cluster is rebalanced.
Hover to see nodes names; edges to Self not shown, Caped at 50 nodes.
Using a canvas is more power efficient and can get hundred of nodes ; but does not allow hyperlinks; , arrows or text (beyond on hover)
SVG is more flexible but power hungry; and does not scale well to 50 + nodes.
All aboves nodes referred to, (or are referred from) current nodes; Edges from Self to other have been omitted (or all nodes would be connected to the central node "self" which is not useful). Nodes are colored by the library they belong to, and scaled with the number of references pointing them