View Source wpool_pool (worker_pool v6.4.0)

Top supervisor for a worker_pool.

This supervisor supervises wpool_process_sup (which is the worker's supervisor) together with auxiliary servers that help keep the whole pool running and in order.

The strategy of this supervisor must be one_for_all but the intensity and period may be changed from their defaults by the t:wpool:pool_sup_intensity() and t:wpool:pool_sup_intensity() options respectively.

Summary

Functions

Adds a callback module. The module must implement the wpool_process_callbacks behaviour.

Picks the worker with the smaller queue of messages.

Calls all workers in the pool in parallel

Casts a message to all the workers within the given pool.

Picks the first available worker and sends the call to it. The timeout provided includes the time it takes to get a worker and for it to process the call.

Casts a message to the first available worker. Since we can wait forever for a wpool:cast to be delivered but we don't want the caller to be blocked, this function just forwards the cast when it gets the worker

Use this function to get the Worker pool record in a custom worker.

Retrieves the list of worker registered names. This can be useful to manually inspect the workers or do custom work on them.

Picks a worker base on a hash result.

Set next within the worker pool record. Useful when using a custom strategy function.

Picks the first available worker, if any

Picks the next worker in a round robin fashion

Picks a random worker

Removes a callback module.

Picks the first available worker and sends the call to it. The timeout provided includes the time it takes to get a worker and for it to process the call.

Picks the first available worker and sends the request to it. The timeout provided considers only the time it takes to get a worker

Starts a supervisor with several wpool_processes as its children

Retrieves the pool stats for all pools

Retrieves a snapshot of the pool stats

Get values from the worker pool record. Useful when using a custom strategy function.

Types

-opaque wpool()

Functions

Link to this function

add_callback_module(Pool, Module)

View Source
-spec add_callback_module(wpool:name(), module()) -> ok | {error, term()}.

Adds a callback module. The module must implement the wpool_process_callbacks behaviour.

-spec best_worker(wpool:name()) -> atom().

Picks the worker with the smaller queue of messages.

Link to this function

broadcall(Name, Call, Timeout)

View Source
-spec broadcall(wpool:name(), term(), timeout()) -> {[Replies :: term()], [Errors :: term()]}.

Calls all workers in the pool in parallel

Waits for responses in parallel too, and it assumes that if any response times out, all of them did too and therefore exits with reason timeout like a regular gen_server does.

-spec broadcast(wpool:name(), term()) -> ok.

Casts a message to all the workers within the given pool.

Link to this function

call_available_worker(Name, Call, Timeout)

View Source
-spec call_available_worker(wpool:name(), any(), timeout()) -> any().

Picks the first available worker and sends the call to it. The timeout provided includes the time it takes to get a worker and for it to process the call.

Link to this function

cast_to_available_worker(Name, Cast)

View Source
-spec cast_to_available_worker(wpool:name(), term()) -> ok.

Casts a message to the first available worker. Since we can wait forever for a wpool:cast to be delivered but we don't want the caller to be blocked, this function just forwards the cast when it gets the worker

-spec find_wpool(atom()) -> undefined | wpool().

Use this function to get the Worker pool record in a custom worker.

-spec get_workers(wpool:name()) -> [atom()].

Retrieves the list of worker registered names. This can be useful to manually inspect the workers or do custom work on them.

Link to this function

hash_worker(Name, HashKey)

View Source
-spec hash_worker(wpool:name(), term()) -> atom().

Picks a worker base on a hash result.

phash2(Term, Range)

returns hash = integer, 0 <= hash < Range so

1

must be added

-spec next(pos_integer(), wpool()) -> wpool().

Set next within the worker pool record. Useful when using a custom strategy function.

Link to this function

next_available_worker(Name)

View Source
-spec next_available_worker(wpool:name()) -> atom().

Picks the first available worker, if any

-spec next_worker(wpool:name()) -> atom().

Picks the next worker in a round robin fashion

-spec random_worker(wpool:name()) -> atom().

Picks a random worker

Link to this function

remove_callback_module(Pool, Module)

View Source
-spec remove_callback_module(wpool:name(), module()) -> ok | {error, term()}.

Removes a callback module.

Link to this function

run_with_available_worker(Name, Run, Timeout)

View Source
-spec run_with_available_worker(wpool:name(), wpool:run(Result), timeout()) -> Result.

Picks the first available worker and sends the call to it. The timeout provided includes the time it takes to get a worker and for it to process the call.

Link to this function

send_request_available_worker(Name, Call, Timeout)

View Source
-spec send_request_available_worker(wpool:name(), any(), timeout()) ->
                                       noproc | timeout | gen_server:request_id().

Picks the first available worker and sends the request to it. The timeout provided considers only the time it takes to get a worker

Link to this function

start_link(Name, Options)

View Source
-spec start_link(wpool:name(), wpool:options()) -> supervisor:startlink_ret().

Starts a supervisor with several wpool_processes as its children

-spec stats() -> [wpool:stats()].

Retrieves the pool stats for all pools

-spec stats(wpool:name()) -> wpool:stats().

Retrieves a snapshot of the pool stats

-spec worker_name(wpool:name(), pos_integer()) -> atom().
-spec wpool_get(atom(), wpool()) -> any();
               ([atom()], wpool()) -> any().

Get values from the worker pool record. Useful when using a custom strategy function.