/* Part of SWI-Prolog Author: Jan Wielemaker E-mail: jan@swi-prolog.org WWW: http://www.swi-prolog.org Copyright (c) 2021, SWI-Prolog Solutions b.v. All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ :- module(stress_client, []). :- use_module(library(stomp)). :- use_module(library(main)). :- use_module(library(debug)). :- use_module(library(option)). :- use_module(library(thread)). :- initialization(main, main). /** Stress testing client This library, together with stress_server.pl and ./server-loop.sh implements stress testing reliable communication. The server is a simple echo server. The script `server-loop.sh` kills the server every 0.2 seconds. The test starts one or more instances of `server-loop.sh`, simulating redundant frequently crashing servers. This clients creates one or more threads that connect to the broker and send a series of messages to the `stress.requests` queue. It listens on the `stress.replies` queue and clears every message for which it receives a reply. At the end it reports teh number of messages it sent and for which it got no reply (this should not happen). Running this occasionally prints "Unexpected (multiple?) replies for ..". That is fine. If a server is killed between replying and acknowledging the reply the request is redelivered and a second reply comes in. To run the stress test: - Make sure a STOMP server (tested RabbitMQ) runs at port 32772 on `localhost` and is accessible as user `guest`, password `guest`. - In three terminals start `./server-loop.sh` - In a fourth terminal run e.g. swipl stress_client.pl --threads=8 --count=100000 */ % configure debugging levels :- set_prolog_flag(message_context, [thread,time]), % debug(stress(send)), % debug(stress(receive)), debug(stress(progress)), debug(stress(wait)), debug(stomp(connection)), true. :- dynamic sent/3. %! main(+Argv) % % Run the test client. Options processed: % % - ``--threads=N``
% Create N client threads. % - ``--count=N``
% Each thread sends N messages. % - ``--tmon``
% Run the XPCE _thread monitor_ during the test and fall back to % the toplevel after completion. % - ``--profile``
% Run profile/1 for the first client. main(Argv) :- argv_options(Argv, _Positional, Options), ( option(tmon(true), Options) -> prolog_ide(thread_monitor) ; true ), option(threads(NThreads), Options, 1), length(Done, NThreads), set_flag(received, 0), set_flag(client, 0), concurrent_maplist(test_client_(Options), Done), ( option(tmon(true), Options) -> cli_enable_development_system ; true ). test_client_(Options, Pass) :- option(profile(true), Options), flag(client, 0, 1), !, profile(test_client(Pass, Options)). test_client_(Options, Pass) :- test_client(Pass, Options). test_client(Pass, Options) :- connect(Connection), option(count(Count), Options, 10), thread_id(Id), requests(Connection, Id, Count), debug(stress(progress), 'All sent, waiting for final replies', []), ( wait_receive(Id) -> Pass = true ; aggregate_all(count, sent(Id, _, _), Missed), format('Thread ~p: no answer: for ~D requests~n', [Id, Missed]), Pass = false ), debug(stress(progress), 'All done', []), stomp_disconnect(Connection, _{}), stomp_destroy_connection(Connection). requests(Connection, Id, N) :- forall(between(1, N, I), send_msg(Connection, Id, I)). send_msg(Connection, Id, Content) :- stomp_send_json(Connection, '/queue/stress.requests', _{}, Tag{sender:Id, content:Content}), get_time(Now), debug(stress(send), 'Sent ~p', [Tag{sender:Id, content:Content}]), assertz(sent(Id, Content, Now)). %! wait_receive(+Id) is semidet. % % Wait for all replies for client Id to be received and succeed if we % got all of them. This uses thread_wait/2 to watch for sent/3 to % drain. After a 10 seconds timeout we check whether any messages are % still being received and continue waiting if this is the case. % % In other words, if not all replies have been received and no new % reply was received for 10 seconds we assume the reply was lost. wait_receive(Id) :- get_flag(received, R0), ( thread_wait(\+ sent(Id,_,_), [ wait_preds([sent/3]), timeout(10) ]) -> debug(stress(wait), '[~w]: received all', [Id]) ; get_flag(received, R1), R1 > R0 -> Rec is R1-R0, debug(stress(wait), '[~w]: received ~D~n', [Id, Rec]), wait_receive(Id) ; debug(stress(wait), '[~w]: received none~n', [Id]), fail ). connect(Connection) :- stomp_connection('127.0.0.1':32772, '/', _{'heart-beat': '5000,5000', login: guest, passcode: guest }, on_frame, Connection, [ reconnect(true), json_options([ value_string_as(atom) ]) ]), stomp_connect(Connection). on_frame(connected, Connection, _Header, _Body) :- debug(pong(connection), 'Extablised connection', []), stomp_subscribe(Connection, '/queue/stress.replies', 0, _{ack: 'client-individual'}). on_frame(disconnected, _Connection, _Header, _Body) :- debug(pong(connection), 'Lost connection', []). on_frame(heartbeat_timeout, _Connection, _Header, _Body) :- debug(pong(connection), 'heartbeat timeout', []). on_frame(heartbeat, _Connection, _Header, _Body) :- debug(pong(heartbeat), 'heartbeat', []). on_frame(error, _Connection, Header, Body) :- print_message(error, stomp_error(Header, Body)). on_frame(message, Connection, Header, _{sender:Id, content:Content}) :- get_time(Now), received(Id, Content, Now), stomp_ack(Connection, Header). received(Id, Content, Received) :- flag(received, N, N+1), ( N mod 1000 =:= 0 -> thread_id(Me), format(user_error, '\r[~w] Received ~D', [Me, N]) ; true ), ( retract(sent(Id, Content, Sent)) -> Time is Received-Sent, debug(stress(received), 'Got ~p in ~6f sec', [#{sender:Id, content:Content}, Time]) ; print_message(warning, not_sent(Id, Content)) ). thread_id(Id) :- thread_self(Me), ( atom(Me) -> Id = Me ; thread_property(Me, id(Id)) ). :- multifile prolog:message//1. prolog:message(not_sent(Id, Content)) --> [ 'Unexpected (multiple?) replies for ~p ~p'-[Id, Content] ].