[ViewVC] Diff of: cvs/AnyEvent-Fork-Pool/Pool.pm

Comparing AnyEvent-Fork-Pool/Pool.pm (file contents):
Revision 1.1 by root, Thu Apr 18 14:24:01 2013 UTC vs.
Revision 1.6 by root, Sun Apr 21 11:17:02 2013 UTC

…		…
6		6
7	use AnyEvent;	7	use AnyEvent;
8	use AnyEvent::Fork::Pool;	8	use AnyEvent::Fork::Pool;
9	# use AnyEvent::Fork is not needed	9	# use AnyEvent::Fork is not needed
10		10
11	# all parameters with default values	11	# all possible parameters shown, with default values
12	my $pool = new AnyEvent::Fork::Pool	12	my $pool = AnyEvent::Fork
13	"MyWorker::run",	13	->new
		14	->require ("MyWorker")
		15	->AnyEvent::Fork::Pool::run (
		16	"MyWorker::run", # the worker function
14		17
15	# pool management	18	# pool management
16	min => 0, # minimum # of processes
17	max => 8, # maximum # of processes	19	max => 4, # absolute maximum # of processes
		20	idle => 0, # minimum # of idle processes
		21	load => 2, # queue at most this number of jobs per process
18	busy_time => 0, # wait this before starting a new process	22	start => 0.1, # wait this many seconds before starting a new process
19	max_idle => 1, # wait this before killing an idle process	23	stop => 10, # wait this many seconds before stopping an idle process
20	idle_time => 1, # at most this many idle processes	24	on_destroy => (my $finish = AE::cv), # called when object is destroyed
21		25
22	# template process
23	template => AnyEvent::Fork->new, # the template process to use
24	require => [MyWorker::], # module(s) to load
25	eval => "# perl code to execute in template",
26	on_destroy => (my $finish = AE::cv),
27
28	# parameters passed to AnyEvent::Fork::RPC	26	# parameters passed to AnyEvent::Fork::RPC
29	async => 0,	27	async => 0,
30	on_error => sub { die "FATAL: $_[0]\n" },	28	on_error => sub { die "FATAL: $_[0]\n" },
31	on_event => sub { my @ev = @_ },	29	on_event => sub { my @ev = @_ },
32	init => "MyWorker::init",	30	init => "MyWorker::init",
33	serialiser => $AnyEvent::Fork::RPC::STRING_SERIALISER,	31	serialiser => $AnyEvent::Fork::RPC::STRING_SERIALISER,
34	;	32	);
35		33
36	for (1..10) {	34	for (1..10) {
37	$pool->call (doit => $_, sub {	35	$pool->(doit => $_, sub {
38	print "MyWorker::run returned @_\n";	36	print "MyWorker::run returned @_\n";
39	});	37	});
40	}	38	}
41		39
42	undef $pool;	40	undef $pool;
…		…
50	pool of processes that handles jobs.	48	pool of processes that handles jobs.
51		49
52	Understanding of L<AnyEvent::Fork> is helpful but not critical to be able	50	Understanding of L<AnyEvent::Fork> is helpful but not critical to be able
53	to use this module, but a thorough understanding of L<AnyEvent::Fork::RPC>	51	to use this module, but a thorough understanding of L<AnyEvent::Fork::RPC>
54	is, as it defines the actual API that needs to be implemented in the	52	is, as it defines the actual API that needs to be implemented in the
55	children.	53	worker processes.
56		54
57	=head1 EXAMPLES	55	=head1 EXAMPLES
58		56
59	=head1 API	57	=head1 PARENT USAGE
		58
		59	To create a pool, you first have to create a L<AnyEvent::Fork> object -
		60	this object becomes your template process. Whenever a new worker process
		61	is needed, it is forked from this template process. Then you need to
		62	"hand off" this template process to the C<AnyEvent::Fork::Pool> module by
		63	calling its run method on it:
		64
		65	my $template = AnyEvent::Fork
		66	->new
		67	->require ("SomeModule", "MyWorkerModule");
		68
		69	my $pool = $template->AnyEvent::Fork::Pool::run ("MyWorkerModule::myfunction");
		70
		71	The pool "object" is not a regular Perl object, but a code reference that
		72	you can call and that works roughly like calling the worker function
		73	directly, except that it returns nothing but instead you need to specify a
		74	callback to be invoked once results are in:
		75
		76	$pool->(1, 2, 3, sub { warn "myfunction(1,2,3) returned @_" });
60		77
61	=over 4	78	=over 4
62		79
63	=cut	80	=cut
64		81
65	package AnyEvent::Fork::Pool;	82	package AnyEvent::Fork::Pool;
66		83
67	use common::sense;	84	use common::sense;
68		85
		86	use Scalar::Util ();
		87
69	use Guard ();	88	use Guard ();
		89	use Array::Heap ();
70		90
71	use AnyEvent;	91	use AnyEvent;
72	use AnyEvent::Fork; # we don't actually depend on it, this is for convenience	92	use AnyEvent::Fork; # we don't actually depend on it, this is for convenience
73	use AnyEvent::Fork::RPC;	93	use AnyEvent::Fork::RPC;
74		94
		95	# these are used for the first and last argument of events
		96	# in the hope of not colliding. yes, I don't like it either,
		97	# but didn't come up with an obviously better alternative.
		98	my $magic0 = ':t6Z@HK1N%Dx@_7?=~-7NQgWDdAs6a,jFN=wLO0jD1%P';
		99	my $magic1 = '<~53rexz.U`!]X[A235^"fyEoiTF\T~oH1l/N6+Djep9b~bI9`\1x%B~vWO1q*';
		100
75	our $VERSION = 0.1;	101	our $VERSION = 0.1;
76		102
77	=item my $rpc = new AnyEvent::Fork::RPC::pool $function, [key => value...]	103	=item my $pool = AnyEvent::Fork::Pool::run $fork, $function, [key => value...]
		104
		105	The traditional way to call the pool creation function. But it is way
		106	cooler to call it in the following way:
		107
		108	=item my $pool = $fork->AnyEvent::Fork::Pool::run ($function, [key => value...])
		109
		110	Creates a new pool object with the specified C<$function> as function
		111	(name) to call for each request. The pool uses the C<$fork> object as the
		112	template when creating worker processes.
		113
		114	You can supply your own template process, or tell C<AnyEvent::Fork::Pool>
		115	to create one.
		116
		117	A relatively large number of key/value pairs can be specified to influence
		118	the behaviour. They are grouped into the categories "pool management",
		119	"template process" and "rpc parameters".
78		120
79	=over 4	121	=over 4
80		122
81	=item on_error => $cb->($msg)	123	=item Pool Management
82		124
83	Called on (fatal) errors, with a descriptive (hopefully) message. If	125	The pool consists of a certain number of worker processes. These options
84	this callback is not provided, but C<on_event> is, then the C<on_event>	126	decide how many of these processes exist and when they are started and
85	callback is called with the first argument being the string C<error>,	127	stopped.
86	followed by the error message.
87		128
88	If neither handler is provided it prints the error to STDERR and will	129	The worker pool is dynamically resized, according to (perceived :)
89	start failing badly.	130	load. The minimum size is given by the C<idle> parameter and the maximum
		131	size is given by the C<max> parameter. A new worker is started every
		132	C<start> seconds at most, and an idle worker is stopped at most every
		133	C<stop> second.
90		134
91	=item on_event => $cb->(...)	135	You can specify the amount of jobs sent to a worker concurrently using the
		136	C<load> parameter.
92		137
93	Called for every call to the C<AnyEvent::Fork::RPC::event> function in the	138	=over 4
94	child, with the arguments of that function passed to the callback.
95		139
96	Also called on errors when no C<on_error> handler is provided.	140	=item idle => $count (default: 0)
97		141
98	=item on_destroy => $cb->()	142	The minimum amount of idle processes in the pool - when there are fewer
		143	than this many idle workers, C<AnyEvent::Fork::Pool> will try to start new
		144	ones, subject to the limits set by C<max> and C<start>.
99		145
100	Called when the C<$rpc> object has been destroyed and all requests have	146	This is also the initial amount of workers in the pool. The default of
101	been successfully handled. This is useful when you queue some requests and	147	zero means that the pool starts empty and can shrink back to zero workers
102	want the child to go away after it has handled them. The problem is that	148	over time.
103	the parent must not exit either until all requests have been handled, and
104	this can be accomplished by waiting for this callback.
105		149
106	=item init => $function (default none)	150	=item max => $count (default: 4)
107		151
108	When specified (by name), this function is called in the child as the very	152	The maximum number of processes in the pool, in addition to the template
109	first thing when taking over the process, with all the arguments normally	153	process. C<AnyEvent::Fork::Pool> will never have more than this number of
110	passed to the C<AnyEvent::Fork::run> function, except the communications	154	worker processes, although there can be more temporarily when a worker is
111	socket.	155	shut down and hasn't exited yet.
112		156
113	It can be used to do one-time things in the child such as storing passed	157	=item load => $count (default: 2)
114	parameters or opening database connections.
115		158
116	It is called very early - before the serialisers are created or the	159	The maximum number of concurrent jobs sent to a single worker process.
117	C<$function> name is resolved into a function reference, so it could be	160
118	used to load any modules that provide the serialiser or function. It can	161	Jobs that cannot be sent to a worker immediately (because all workers are
119	not, however, create events.	162	busy) will be queued until a worker is available.
		163
		164	Setting this low improves latency. For example, at C<1>, every job that
		165	is sent to a worker is sent to a completely idle worker that doesn't run
		166	any other jobs. The downside is that throughput is reduced - a worker that
		167	finishes a job needs to wait for a new job from the parent.
		168
		169	The default of C<2> is usually a good compromise.
		170
		171	=item start => $seconds (default: 0.1)
		172
		173	When there are fewer than C<idle> workers (or all workers are completely
		174	busy), then a timer is started. If the timer elapses and there are still
		175	jobs that cannot be queued to a worker, a new worker is started.
		176
		177	This sets the minimum time that all workers must be busy before a new
		178	worker is started. Or, put differently, the minimum delay between starting
		179	new workers.
		180
		181	The delay is small by default, which means new workers will be started
		182	relatively quickly. A delay of C<0> is possible, and ensures that the pool
		183	will grow as quickly as possible under load.
		184
		185	Non-zero values are useful to avoid "exploding" a pool because a lot of
		186	jobs are queued in an instant.
		187
		188	Higher values are often useful to improve efficiency at the cost of
		189	latency - when fewer processes can do the job over time, starting more and
		190	more is not necessarily going to help.
		191
		192	=item stop => $seconds (default: 10)
		193
		194	When a worker has no jobs to execute it becomes idle. An idle worker that
		195	hasn't executed a job within this amount of time will be stopped, unless
		196	the other parameters say otherwise.
		197
		198	Setting this to a very high value means that workers stay around longer,
		199	even when they have nothing to do, which can be good as they don't have to
		200	be started on the netx load spike again.
		201
		202	Setting this to a lower value can be useful to avoid memory or simply
		203	process table wastage.
		204
		205	Usually, setting this to a time longer than the time between load spikes
		206	is best - if you expect a lot of requests every minute and little work
		207	in between, setting this to longer than a minute avoids having to stop
		208	and start workers. On the other hand, you have to ask yourself if letting
		209	workers run idle is a good use of your resources. Try to find a good
		210	balance between resource usage of your workers and the time to start new
		211	workers - the processes created by L<AnyEvent::Fork> itself is fats at
		212	creating workers while not using much memory for them, so most of the
		213	overhead is likely from your own code.
		214
		215	=item on_destroy => $callback->() (default: none)
		216
		217	When a pool object goes out of scope, the outstanding requests are still
		218	handled till completion. Only after handling all jobs will the workers
		219	be destroyed (and also the template process if it isn't referenced
		220	otherwise).
		221
		222	To find out when a pool I<really> has finished its work, you can set this
		223	callback, which will be called when the pool has been destroyed.
		224
		225	=back
		226
		227	=item AnyEvent::Fork::RPC Parameters
		228
		229	These parameters are all passed more or less directly to
		230	L<AnyEvent::Fork::RPC>. They are only briefly mentioned here, for
		231	their full documentation please refer to the L<AnyEvent::Fork::RPC>
		232	documentation. Also, the default values mentioned here are only documented
		233	as a best effort - the L<AnyEvent::Fork::RPC> documentation is binding.
		234
		235	=over 4
120		236
121	=item async => $boolean (default: 0)	237	=item async => $boolean (default: 0)
122		238
123	The default server used in the child does all I/O blockingly, and only	239	Whether to use the synchronous or asynchronous RPC backend.
124	allows a single RPC call to execute concurrently.
125		240
126	Setting C<async> to a true value switches to another implementation that	241	=item on_error => $callback->($message) (default: die with message)
127	uses L<AnyEvent> in the child and allows multiple concurrent RPC calls.
128		242
129	The actual API in the child is documented in the section that describes	243	The callback to call on any (fatal) errors.
130	the calling semantics of the returned C<$rpc> function.
131		244
132	If you want to pre-load the actual back-end modules to enable memory	245	=item on_event => $callback->(...) (default: C<sub { }>, unlike L<AnyEvent::Fork::RPC>)
133	sharing, then you should load C<AnyEvent::Fork::RPC::Sync> for
134	synchronous, and C<AnyEvent::Fork::RPC::Async> for asynchronous mode.
135		246
136	If you use a template process and want to fork both sync and async	247	The callback to invoke on events.
137	children, then it is permissible to load both modules.
138		248
139	=item serialiser => $string (default: '(sub { pack "(w/a)", @_ }, sub { unpack "(w/a)", shift })')	249	=item init => $initfunction (default: none)
140		250
141	All arguments, result data and event data have to be serialised to be	251	The function to call in the child, once before handling requests.
142	transferred between the processes. For this, they have to be frozen and
143	thawed in both parent and child processes.
144		252
145	By default, only octet strings can be passed between the processes, which	253	=item serialiser => $serialiser (defailt: $AnyEvent::Fork::RPC::STRING_SERIALISER)
146	is reasonably fast and efficient.
147		254
148	For more complicated use cases, you can provide your own freeze and thaw	255	The serialiser to use.
149	functions, by specifying a string with perl source code. It's supposed to
150	return two code references when evaluated: the first receives a list of
151	perl values and must return an octet string. The second receives the octet
152	string and must return the original list of values.
153
154	If you need an external module for serialisation, then you can either
155	pre-load it into your L<AnyEvent::Fork> process, or you can add a C<use>
156	or C<require> statement into the serialiser string. Or both.
157		256
158	=back	257	=back
159		258
160	See the examples section earlier in this document for some actual	259	=back
161	examples.
162		260
163	=cut	261	=cut
164		262
165	sub new {	263	sub run {
166	my ($self, $function, %arg) = @_;	264	my ($template, $function, %arg) = @_;
167		265
168	my $serialiser = delete $arg{serialiser} \|\| $STRING_SERIALISER;	266	my $max = $arg{max} \|\| 4;
		267	my $idle = $arg{idle} \|\| 0,
		268	my $load = $arg{load} \|\| 2,
		269	my $start = $arg{start} \|\| 0.1,
		270	my $stop = $arg{stop} \|\| 10,
169	my $on_event = delete $arg{on_event};	271	my $on_event = $arg{on_event} \|\| sub { },
170	my $on_error = delete $arg{on_error};
171	my $on_destroy = delete $arg{on_destroy};	272	my $on_destroy = $arg{on_destroy};
		273
		274	my @rpc = (
		275	async => $arg{async},
		276	init => $arg{init},
		277	serialiser => delete $arg{serialiser},
		278	on_error => $arg{on_error},
172		279	);
173	# default for on_error is to on_event, if specified
174	$on_error \|\|= $on_event
175	? sub { $on_event->(error => shift) }
176	: sub { die "AnyEvent::Fork::RPC: uncaught error: $_[0].\n" };
177		280
178	# default for on_event is to raise an error	281	my (@pool, @queue, $nidle, $start_w, $stop_w, $shutdown);
179	$on_event \|\|= sub { $on_error->("event received, but no on_event handler") };	282	my ($start_worker, $stop_worker, $want_start, $want_stop, $scheduler);
180		283
181	my ($f, $t) = eval $serialiser; die $@ if $@;	284	my $destroy_guard = Guard::guard {
		285	$on_destroy->()
		286	if $on_destroy;
		287	};
182		288
183	my (@rcb, %rcb, $fh, $shutdown, $wbuf, $ww);	289	$template
184	my ($rlen, $rbuf, $rw) = 512 - 16;	290	->require ("AnyEvent::Fork::RPC::" . ($arg{async} ? "Async" : "Sync"))
185		291	->eval ('
186	my $wcb = sub {	292	my ($magic0, $magic1) = @_;
187	my $len = syswrite $fh, $wbuf;	293	sub AnyEvent::Fork::Pool::retire() {
188		294	AnyEvent::Fork::RPC::event $magic0, "quit", $magic1;
189	unless (defined $len) {
190	if ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) {
191	undef $rw; undef $ww; # it ends here
192	$on_error->("$!");
193	}	295	}
194	}	296	', $magic0, $magic1)
195
196	substr $wbuf, 0, $len, "";
197
198	unless (length $wbuf) {
199	undef $ww;
200	$shutdown and shutdown $fh, 1;
201	}
202	};	297	;
203		298
204	my $module = "AnyEvent::Fork::RPC::" . ($arg{async} ? "Async" : "Sync");	299	$start_worker = sub {
		300	my $proc = [0, 0, undef]; # load, index, rpc
205		301
206	$self->require ($module)	302	$proc->[2] = $template
207	->send_arg ($function, $arg{init}, $serialiser)	303	->fork
208	->run ("$module\::run", sub {	304	->AnyEvent::Fork::RPC::run ($function,
209	$fh = shift;	305	@rpc,
		306	on_event => sub {
		307	if (@_ == 3 && $_[0] eq $magic0 && $_[2] eq $magic1) {
		308	$destroy_guard if 0; # keep it alive
210		309
211	my ($id, $len);	310	$_[1] eq "quit" and $stop_worker->($proc);
212	$rw = AE::io $fh, 0, sub {
213	$rlen = $rlen * 2 + 16 if $rlen - 128 < length $rbuf;
214	$len = sysread $fh, $rbuf, $rlen - length $rbuf, length $rbuf;
215
216	if ($len) {
217	while (8 <= length $rbuf) {
218	($id, $len) = unpack "LL", $rbuf;
219	8 + $len <= length $rbuf
220	or last;	311	return;
221
222	my @r = $t->(substr $rbuf, 8, $len);
223	substr $rbuf, 0, 8 + $len, "";
224
225	if ($id) {
226	if (@rcb) {
227	(shift @rcb)->(@r);
228	} elsif (my $cb = delete $rcb{$id}) {
229	$cb->(@r);
230	} else {
231	undef $rw; undef $ww;
232	$on_error->("unexpected data from child");
233	}	312	}
234	} else {	313
235	$on_event->(@r);	314	&$on_event;
236	}	315	},
237	}	316	)
238	} elsif (defined $len) {	317	;
239	undef $rw; undef $ww; # it ends here
240		318
241	if (@rcb \|\| %rcb) {	319	++$nidle;
242	$on_error->("unexpected eof");	320	Array::Heap::push_heap_idx @pool, $proc;
		321
		322	Scalar::Util::weaken $proc;
		323	};
		324
		325	$stop_worker = sub {
		326	my $proc = shift;
		327
		328	$proc->[0]
		329	or --$nidle;
		330
		331	Array::Heap::splice_heap_idx @pool, $proc->[1]
		332	if defined $proc->[1];
		333	};
		334
		335	$want_start = sub {
		336	undef $stop_w;
		337
		338	$start_w \|\|= AE::timer $start, $start, sub {
		339	if (($nidle < $idle \|\| @queue) && @pool < $max) {
		340	$start_worker->();
		341	$scheduler->();
243	} else {	342	} else {
244	$on_destroy->();	343	undef $start_w;
245	}
246	} elsif ($! != Errno::EAGAIN && $! != Errno::EWOULDBLOCK) {
247	undef $rw; undef $ww; # it ends here
248	$on_error->("read: $!");
249	}	344	}
250	};	345	};
251
252	$ww \|\|= AE::io $fh, 1, $wcb;
253	});	346	};
254		347
		348	$want_stop = sub {
		349	$stop_w \|\|= AE::timer $stop, $stop, sub {
		350	$stop_worker->($pool[0])
		351	if $nidle;
		352
		353	undef $stop_w
		354	if $nidle <= $idle;
		355	};
		356	};
		357
		358	$scheduler = sub {
		359	if (@queue) {
		360	while (@queue) {
		361	my $proc = $pool[0];
		362
		363	if ($proc->[0] < $load) {
		364	# found free worker, increase load
		365	unless ($proc->[0]++) {
		366	# worker became busy
		367	--$nidle
		368	or undef $stop_w;
		369
		370	$want_start->()
		371	if $nidle < $idle && @pool < $max;
		372	}
		373
		374	Array::Heap::adjust_heap_idx @pool, 0;
		375
		376	my $job = shift @queue;
		377	my $ocb = pop @$job;
		378
		379	$proc->[2]->(@$job, sub {
		380	# reduce load
		381	--$proc->[0] # worker still busy?
		382	or ++$nidle > $idle # not too many idle processes?
		383	or $want_stop->();
		384
		385	Array::Heap::adjust_heap_idx @pool, $proc->[1]
		386	if defined $proc->[1];
		387
		388	$scheduler->();
		389
		390	&$ocb;
		391	});
		392	} else {
		393	$want_start->()
		394	unless @pool >= $max;
		395
		396	last;
		397	}
		398	}
		399	} elsif ($shutdown) {
		400	@pool = ();
		401	undef $start_w;
		402	undef $start_worker; # frees $destroy_guard reference
		403
		404	$stop_worker->($pool[0])
		405	while $nidle;
		406	}
		407	};
		408
255	my $guard = Guard::guard {	409	my $shutdown_guard = Guard::guard {
256	$shutdown = 1;	410	$shutdown = 1;
257	$ww \|\|= $fh && AE::io $fh, 1, $wcb;	411	$scheduler->();
		412	};
		413
		414	$start_worker->()
		415	while @pool < $idle;
		416
		417	sub {
		418	$shutdown_guard if 0; # keep it alive
		419
		420	$start_worker->()
		421	unless @pool;
		422
		423	push @queue, [@_];
		424	$scheduler->();
258	};	425	}
259
260	my $id;
261
262	$arg{async}
263	? sub {
264	$id = ($id == 0xffffffff ? 0 : $id) + 1;
265	$id = ($id == 0xffffffff ? 0 : $id) + 1 while exists $rcb{$id}; # rarely loops
266
267	$rcb{$id} = pop;
268
269	$guard; # keep it alive
270
271	$wbuf .= pack "LL/a*", $id, &$f;
272	$ww \|\|= $fh && AE::io $fh, 1, $wcb;
273	}
274	: sub {
275	push @rcb, pop;
276
277	$guard; # keep it alive
278
279	$wbuf .= pack "L/a*", &$f;
280	$ww \|\|= $fh && AE::io $fh, 1, $wcb;
281	}
282	}	426	}
283		427
284	=item $pool->call (..., $cb->(...))	428	=item $pool->(..., $cb->(...))
		429
		430	Call the RPC function of a worker with the given arguments, and when the
		431	worker is done, call the C<$cb> with the results, just like calling the
		432	RPC object durectly - see the L<AnyEvent::Fork::RPC> documentation for
		433	details on the RPC API.
		434
		435	If there is no free worker, the call will be queued until a worker becomes
		436	available.
		437
		438	Note that there can be considerable time between calling this method and
		439	the call actually being executed. During this time, the parameters passed
		440	to this function are effectively read-only - modifying them after the call
		441	and before the callback is invoked causes undefined behaviour.
		442
		443	=cut
		444
		445	=back
		446
		447	=head1 CHILD USAGE
		448
		449	In addition to the L<AnyEvent::Fork::RPC> API, this module implements one
		450	more child-side function:
		451
		452	=over 4
		453
		454	=item AnyEvent::Fork::Pool::retire ()
		455
		456	This function sends an event to the parent process to request retirement:
		457	the worker is removed from the pool and no new jobs will be sent to it,
		458	but it has to handle the jobs that are already queued.
		459
		460	The parentheses are part of the syntax: the function usually isn't defined
		461	when you compile your code (because that happens I<before> handing the
		462	template process over to C<AnyEvent::Fork::Pool::run>, so you need the
		463	empty parentheses to tell Perl that the function is indeed a function.
		464
		465	Retiring a worker can be useful to gracefully shut it down when the worker
		466	deems this useful. For example, after executing a job, one could check
		467	the process size or the number of jobs handled so far, and if either is
		468	too high, the worker could ask to get retired, to avoid memory leaks to
		469	accumulate.
		470
		471	=back
		472
		473	=head1 POOL PARAMETERS RECIPES
		474
		475	This section describes some recipes for pool paramaters. These are mostly
		476	meant for the synchronous RPC backend, as the asynchronous RPC backend
		477	changes the rules considerably, making workers themselves responsible for
		478	their scheduling.
		479
		480	=over 4
		481
		482	=item low latency - set load = 1
		483
		484	If you need a deterministic low latency, you should set the C<load>
		485	parameter to C<1>. This ensures that never more than one job is sent to
		486	each worker. This avoids having to wait for a previous job to finish.
		487
		488	This makes most sense with the synchronous (default) backend, as the
		489	asynchronous backend can handle multiple requests concurrently.
		490
		491	=item lowest latency - set load = 1 and idle = max
		492
		493	To achieve the lowest latency, you additionally should disable any dynamic
		494	resizing of the pool by setting C<idle> to the same value as C<max>.
		495
		496	=item high throughput, cpu bound jobs - set load >= 2, max = #cpus
		497
		498	To get high throughput with cpu-bound jobs, you should set the maximum
		499	pool size to the number of cpus in your system, and C<load> to at least
		500	C<2>, to make sure there can be another job waiting for the worker when it
		501	has finished one.
		502
		503	The value of C<2> for C<load> is the minimum value that I<can> achieve
		504	100% throughput, but if your parent process itself is sometimes busy, you
		505	might need higher values. Also there is a limit on the amount of data that
		506	can be "in flight" to the worker, so if you send big blobs of data to your
		507	worker, C<load> might have much less of an effect.
		508
		509	=item high throughput, I/O bound jobs - set load >= 2, max = 1, or very high
		510
		511	When your jobs are I/O bound, using more workers usually boils down to
		512	higher throughput, depending very much on your actual workload - sometimes
		513	having only one worker is best, for example, when you read or write big
		514	files at maixmum speed, as a second worker will increase seek times.
285		515
286	=back	516	=back
287		517
288	=head1 SEE ALSO	518	=head1 SEE ALSO
289		519

Diff Legend

-–
+Removed lines
-+
+Added lines
-<
+Changed lines
->
+Changed lines

Comparing AnyEvent-Fork-Pool/Pool.pm (file contents): Revision 1.1 by root, Thu Apr 18 14:24:01 2013 UTC vs. Revision 1.6 by root, Sun Apr 21 11:17:02 2013 UTC

Diff Legend

Comparing AnyEvent-Fork-Pool/Pool.pm (file contents):
Revision 1.1 by root, Thu Apr 18 14:24:01 2013 UTC vs.
Revision 1.6 by root, Sun Apr 21 11:17:02 2013 UTC