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

Comparing AnyEvent-Fork/Fork.pm (file contents):
Revision 1.24 by root, Sat Apr 6 08:32:23 2013 UTC vs.
Revision 1.37 by root, Sat Apr 6 20:06:39 2013 UTC

…		…
40	in whatever way you like, use some message-passing module such	40	in whatever way you like, use some message-passing module such
41	as L<AnyEvent::MP>, some pipe such as L<AnyEvent::ZeroMQ>, use	41	as L<AnyEvent::MP>, some pipe such as L<AnyEvent::ZeroMQ>, use
42	L<AnyEvent::Handle> on both sides to send e.g. JSON or Storable messages,	42	L<AnyEvent::Handle> on both sides to send e.g. JSON or Storable messages,
43	and so on.	43	and so on.
44		44
		45	=head1 PROBLEM STATEMENT
		46
		47	There are two traditional ways to implement parallel processing on UNIX
		48	like operating systems - fork and process, and fork+exec and process. They
		49	have different advantages and disadvantages that I describe below,
		50	together with how this module tries to mitigate the disadvantages.
		51
		52	=over 4
		53
		54	=item Forking from a big process can be very slow.
		55
		56	A 5GB process needs 0.05s to fork on my 3.6GHz amd64 GNU/Linux box. This
		57	overhead is often shared with exec (because you have to fork first), but
		58	in some circumstances (e.g. when vfork is used), fork+exec can be much
		59	faster.
		60
		61	This module can help here by telling a small(er) helper process to fork,
		62	which is faster then forking the main process, and also uses vfork where
		63	possible. This gives the speed of vfork, with the flexibility of fork.
		64
		65	=item Forking usually creates a copy-on-write copy of the parent
		66	process.
		67
		68	For example, modules or data files that are loaded will not use additional
		69	memory after a fork. When exec'ing a new process, modules and data files
		70	might need to be loaded again, at extra CPU and memory cost. But when
		71	forking, literally all data structures are copied - if the program frees
		72	them and replaces them by new data, the child processes will retain the
		73	old version even if it isn't used, which can suddenly and unexpectedly
		74	increase memory usage when freeing memory.
		75
		76	The trade-off is between more sharing with fork (which can be good or
		77	bad), and no sharing with exec.
		78
		79	This module allows the main program to do a controlled fork, and allows
		80	modules to exec processes safely at any time. When creating a custom
		81	process pool you can take advantage of data sharing via fork without
		82	risking to share large dynamic data structures that will blow up child
		83	memory usage.
		84
		85	In other words, this module puts you into control over what is being
		86	shared and what isn't, at all times.
		87
		88	=item Exec'ing a new perl process might be difficult.
		89
		90	For example, it is not easy to find the correct path to the perl
		91	interpreter - C<$^X> might not be a perl interpreter at all.
		92
		93	This module tries hard to identify the correct path to the perl
		94	interpreter. With a cooperative main program, exec'ing the interpreter
		95	might not even be necessary, but even without help from the main program,
		96	it will still work when used from a module.
		97
		98	=item Exec'ing a new perl process might be slow, as all necessary modules
		99	have to be loaded from disk again, with no guarantees of success.
		100
		101	Long running processes might run into problems when perl is upgraded
		102	and modules are no longer loadable because they refer to a different
		103	perl version, or parts of a distribution are newer than the ones already
		104	loaded.
		105
		106	This module supports creating pre-initialised perl processes to be used as
		107	a template for new processes.
		108
		109	=item Forking might be impossible when a program is running.
		110
		111	For example, POSIX makes it almost impossible to fork from a
		112	multi-threaded program while doing anything useful in the child - in
		113	fact, if your perl program uses POSIX threads (even indirectly via
		114	e.g. L<IO::AIO> or L<threads>), you cannot call fork on the perl level
		115	anymore without risking corruption issues on a number of operating
		116	systems.
		117
		118	This module can safely fork helper processes at any time, by calling
		119	fork+exec in C, in a POSIX-compatible way (via L<Proc::FastSpawn>).
		120
		121	=item Parallel processing with fork might be inconvenient or difficult
		122	to implement. Modules might not work in both parent and child.
		123
		124	For example, when a program uses an event loop and creates watchers it
		125	becomes very hard to use the event loop from a child program, as the
		126	watchers already exist but are only meaningful in the parent. Worse, a
		127	module might want to use such a module, not knowing whether another module
		128	or the main program also does, leading to problems.
		129
		130	Apart from event loops, graphical toolkits also commonly fall into the
		131	"unsafe module" category, or just about anything that communicates with
		132	the external world, such as network libraries and file I/O modules, which
		133	usually don't like being copied and then allowed to continue in two
		134	processes.
		135
		136	With this module only the main program is allowed to create new processes
		137	by forking (because only the main program can know when it is still safe
		138	to do so) - all other processes are created via fork+exec, which makes it
		139	possible to use modules such as event loops or window interfaces safely.
		140
		141	=back
		142
45	=head1 EXAMPLES	143	=head1 EXAMPLES
46		144
47	=head2 Create a single new process, tell it to run your worker function.	145	=head2 Create a single new process, tell it to run your worker function.
48		146
49	AnyEvent::Fork	147	AnyEvent::Fork
…		…
54		152
55	# now $master_filehandle is connected to the	153	# now $master_filehandle is connected to the
56	# $slave_filehandle in the new process.	154	# $slave_filehandle in the new process.
57	});	155	});
58		156
59	# MyModule::worker might look like this	157	C<MyModule> might look like this:
		158
		159	package MyModule;
		160
60	sub MyModule::worker {	161	sub worker {
61	my ($slave_filehandle) = @_;	162	my ($slave_filehandle) = @_;
62		163
63	# now $slave_filehandle is connected to the $master_filehandle	164	# now $slave_filehandle is connected to the $master_filehandle
64	# in the original prorcess. have fun!	165	# in the original prorcess. have fun!
65	}	166	}
…		…
84	}	185	}
85		186
86	# now do other things - maybe use the filehandle provided by run	187	# now do other things - maybe use the filehandle provided by run
87	# to wait for the processes to die. or whatever.	188	# to wait for the processes to die. or whatever.
88		189
89	# My::Server::run might look like this	190	C<My::Server> might look like this:
90	sub My::Server::run {	191
		192	package My::Server;
		193
		194	sub run {
91	my ($slave, $listener, $id) = @_;	195	my ($slave, $listener, $id) = @_;
92		196
93	close $slave; # we do not use the socket, so close it to save resources	197	close $slave; # we do not use the socket, so close it to save resources
94		198
95	# we could go ballistic and use e.g. AnyEvent here, or IO::AIO,	199	# we could go ballistic and use e.g. AnyEvent here, or IO::AIO,
…		…
99	}	203	}
100	}	204	}
101		205
102	=head2 use AnyEvent::Fork as a faster fork+exec	206	=head2 use AnyEvent::Fork as a faster fork+exec
103		207
104	This runs /bin/echo hi, with stdout redirected to /tmp/log and stderr to	208	This runs C</bin/echo hi>, with stdandard output redirected to /tmp/log
105	the communications socket. It is usually faster than fork+exec, but still	209	and standard error redirected to the communications socket. It is usually
106	let's you prepare the environment.	210	faster than fork+exec, but still lets you prepare the environment.
107		211
108	open my $output, ">/tmp/log" or die "$!";	212	open my $output, ">/tmp/log" or die "$!";
109		213
110	AnyEvent::Fork	214	AnyEvent::Fork
111	->new	215	->new
…		…
123	->send_fh ($output)	227	->send_fh ($output)
124	->send_arg ("/bin/echo", "hi")	228	->send_arg ("/bin/echo", "hi")
125	->run ("run", my $cv = AE::cv);	229	->run ("run", my $cv = AE::cv);
126		230
127	my $stderr = $cv->recv;	231	my $stderr = $cv->recv;
128
129	=head1 PROBLEM STATEMENT
130
131	There are two ways to implement parallel processing on UNIX like operating
132	systems - fork and process, and fork+exec and process. They have different
133	advantages and disadvantages that I describe below, together with how this
134	module tries to mitigate the disadvantages.
135
136	=over 4
137
138	=item Forking from a big process can be very slow (a 5GB process needs
139	0.05s to fork on my 3.6GHz amd64 GNU/Linux box for example). This overhead
140	is often shared with exec (because you have to fork first), but in some
141	circumstances (e.g. when vfork is used), fork+exec can be much faster.
142
143	This module can help here by telling a small(er) helper process to fork,
144	or fork+exec instead.
145
146	=item Forking usually creates a copy-on-write copy of the parent
147	process. Memory (for example, modules or data files that have been
148	will not take additional memory). When exec'ing a new process, modules
149	and data files might need to be loaded again, at extra CPU and memory
150	cost. Likewise when forking, all data structures are copied as well - if
151	the program frees them and replaces them by new data, the child processes
152	will retain the memory even if it isn't used.
153
154	This module allows the main program to do a controlled fork, and allows
155	modules to exec processes safely at any time. When creating a custom
156	process pool you can take advantage of data sharing via fork without
157	risking to share large dynamic data structures that will blow up child
158	memory usage.
159
160	=item Exec'ing a new perl process might be difficult and slow. For
161	example, it is not easy to find the correct path to the perl interpreter,
162	and all modules have to be loaded from disk again. Long running processes
163	might run into problems when perl is upgraded for example.
164
165	This module supports creating pre-initialised perl processes to be used
166	as template, and also tries hard to identify the correct path to the perl
167	interpreter. With a cooperative main program, exec'ing the interpreter
168	might not even be necessary.
169
170	=item Forking might be impossible when a program is running. For example,
171	POSIX makes it almost impossible to fork from a multi-threaded program and
172	do anything useful in the child - strictly speaking, if your perl program
173	uses posix threads (even indirectly via e.g. L<IO::AIO> or L<threads>),
174	you cannot call fork on the perl level anymore, at all.
175
176	This module can safely fork helper processes at any time, by calling
177	fork+exec in C, in a POSIX-compatible way.
178
179	=item Parallel processing with fork might be inconvenient or difficult
180	to implement. For example, when a program uses an event loop and creates
181	watchers it becomes very hard to use the event loop from a child
182	program, as the watchers already exist but are only meaningful in the
183	parent. Worse, a module might want to use such a system, not knowing
184	whether another module or the main program also does, leading to problems.
185
186	This module only lets the main program create pools by forking (because
187	only the main program can know when it is still safe to do so) - all other
188	pools are created by fork+exec, after which such modules can again be
189	loaded.
190
191	=back
192		232
193	=head1 CONCEPTS	233	=head1 CONCEPTS
194		234
195	This module can create new processes either by executing a new perl	235	This module can create new processes either by executing a new perl
196	process, or by forking from an existing "template" process.	236	process, or by forking from an existing "template" process.
…		…
275	my ($fork_fh) = @_;	315	my ($fork_fh) = @_;
276	});	316	});
277		317
278	=back	318	=back
279		319
280	=head1 FUNCTIONS	320	=head1 THE C<AnyEvent::Fork> CLASS
		321
		322	This module exports nothing, and only implements a single class -
		323	C<AnyEvent::Fork>.
		324
		325	There are two class constructors that both create new processes - C<new>
		326	and C<new_exec>. The C<fork> method creates a new process by forking an
		327	existing one and could be considered a third constructor.
		328
		329	Most of the remaining methods deal with preparing the new process, by
		330	loading code, evaluating code and sending data to the new process. They
		331	usually return the process object, so you can chain method calls.
		332
		333	If a process object is destroyed before calling its C<run> method, then
		334	the process simply exits. After C<run> is called, all responsibility is
		335	passed to the specified function.
		336
		337	As long as there is any outstanding work to be done, process objects
		338	resist being destroyed, so there is no reason to store them unless you
		339	need them later - configure and forget works just fine.
281		340
282	=over 4	341	=over 4
283		342
284	=cut	343	=cut
285		344
…		…
295	use IO::FDPass;	354	use IO::FDPass;
296		355
297	our $VERSION = 0.5;	356	our $VERSION = 0.5;
298		357
299	our $PERL; # the path to the perl interpreter, deduces with various forms of magic	358	our $PERL; # the path to the perl interpreter, deduces with various forms of magic
300
301	=item my $pool = new AnyEvent::Fork key => value...
302
303	Create a new process pool. The following named parameters are supported:
304		359
305	=over 4	360	=over 4
306		361
307	=back	362	=back
308		363
…		…
404	Create a new "empty" perl interpreter process and returns its process	459	Create a new "empty" perl interpreter process and returns its process
405	object for further manipulation.	460	object for further manipulation.
406		461
407	The new process is forked from a template process that is kept around	462	The new process is forked from a template process that is kept around
408	for this purpose. When it doesn't exist yet, it is created by a call to	463	for this purpose. When it doesn't exist yet, it is created by a call to
409	C<new_exec> and kept around for future calls.	464	C<new_exec> first and then stays around for future calls.
410
411	When the process object is destroyed, it will release the file handle
412	that connects it with the new process. When the new process has not yet
413	called C<run>, then the process will exit. Otherwise, what happens depends
414	entirely on the code that is executed.
415		465
416	=cut	466	=cut
417		467
418	sub new {	468	sub new {
419	my $class = shift;	469	my $class = shift;
…		…
509	}	559	}
510		560
511	=item $pid = $proc->pid	561	=item $pid = $proc->pid
512		562
513	Returns the process id of the process I<iff it is a direct child of the	563	Returns the process id of the process I<iff it is a direct child of the
514	process> running AnyEvent::Fork, and C<undef> otherwise.	564	process running AnyEvent::Fork>, and C<undef> otherwise.
515		565
516	Normally, only processes created via C<< AnyEvent::Fork->new_exec >> and	566	Normally, only processes created via C<< AnyEvent::Fork->new_exec >> and
517	L<AnyEvent::Fork::Template> are direct children, and you are responsible	567	L<AnyEvent::Fork::Template> are direct children, and you are responsible
518	to clean up their zombies when they die.	568	to clean up their zombies when they die.
519		569
520	All other processes are not direct children, and will be cleaned up by	570	All other processes are not direct children, and will be cleaned up by
521	AnyEvent::Fork.	571	AnyEvent::Fork itself.
522		572
523	=cut	573	=cut
524		574
525	sub pid {	575	sub pid {
526	$_[0][0]	576	$_[0][0]
…		…
537		587
538	The code will usually be executed after this call returns, and there is no	588	The code will usually be executed after this call returns, and there is no
539	way to pass anything back to the calling process. Any evaluation errors	589	way to pass anything back to the calling process. Any evaluation errors
540	will be reported to stderr and cause the process to exit.	590	will be reported to stderr and cause the process to exit.
541		591
542	If you want to execute some code to take over the process (see the	592	If you want to execute some code (that isn't in a module) to take over the
543	"fork+exec" example in the SYNOPSIS), you should compile a function via	593	process, you should compile a function via C<eval> first, and then call
544	C<eval> first, and then call it via C<run>. This also gives you access to	594	it via C<run>. This also gives you access to any arguments passed via the
545	any arguments passed via the C<send_xxx> methods, such as file handles.	595	C<send_xxx> methods, such as file handles. See the L<use AnyEvent::Fork as
		596	a faster fork+exec> example to see it in action.
546		597
547	Returns the process object for easy chaining of method calls.	598	Returns the process object for easy chaining of method calls.
548		599
549	=cut	600	=cut
550		601
…		…
576	=item $proc = $proc->send_fh ($handle, ...)	627	=item $proc = $proc->send_fh ($handle, ...)
577		628
578	Send one or more file handles (I<not> file descriptors) to the process,	629	Send one or more file handles (I<not> file descriptors) to the process,
579	to prepare a call to C<run>.	630	to prepare a call to C<run>.
580		631
581	The process object keeps a reference to the handles until this is done,	632	The process object keeps a reference to the handles until they have
582	so you must not explicitly close the handles. This is most easily	633	been passed over to the process, so you must not explicitly close the
583	accomplished by simply not storing the file handles anywhere after passing	634	handles. This is most easily accomplished by simply not storing the file
584	them to this method.	635	handles anywhere after passing them to this method - when AnyEvent::Fork
		636	is finished using them, perl will automatically close them.
585		637
586	Returns the process object for easy chaining of method calls.	638	Returns the process object for easy chaining of method calls.
587		639
588	Example: pass a file handle to a process, and release it without	640	Example: pass a file handle to a process, and release it without
589	closing. It will be closed automatically when it is no longer used.	641	closing. It will be closed automatically when it is no longer used.
…		…
605	}	657	}
606		658
607	=item $proc = $proc->send_arg ($string, ...)	659	=item $proc = $proc->send_arg ($string, ...)
608		660
609	Send one or more argument strings to the process, to prepare a call to	661	Send one or more argument strings to the process, to prepare a call to
610	C<run>. The strings can be any octet string.	662	C<run>. The strings can be any octet strings.
611		663
612	The protocol is optimised to pass a moderate number of relatively short	664	The protocol is optimised to pass a moderate number of relatively short
613	strings - while you can pass up to 4GB of data in one go, this is more	665	strings - while you can pass up to 4GB of data in one go, this is more
614	meant to pass some ID information or other startup info, not big chunks of	666	meant to pass some ID information or other startup info, not big chunks of
615	data.	667	data.
…		…
631	Enter the function specified by the function name in C<$func> in the	683	Enter the function specified by the function name in C<$func> in the
632	process. The function is called with the communication socket as first	684	process. The function is called with the communication socket as first
633	argument, followed by all file handles and string arguments sent earlier	685	argument, followed by all file handles and string arguments sent earlier
634	via C<send_fh> and C<send_arg> methods, in the order they were called.	686	via C<send_fh> and C<send_arg> methods, in the order they were called.
635		687
		688	The process object becomes unusable on return from this function - any
		689	further method calls result in undefined behaviour.
		690
636	The function name should be fully qualified, but if it isn't, it will be	691	The function name should be fully qualified, but if it isn't, it will be
637	looked up in the main package.	692	looked up in the C<main> package.
638		693
639	If the called function returns, doesn't exist, or any error occurs, the	694	If the called function returns, doesn't exist, or any error occurs, the
640	process exits.	695	process exits.
641		696
642	Preparing the process is done in the background - when all commands have	697	Preparing the process is done in the background - when all commands have
643	been sent, the callback is invoked with the local communications socket	698	been sent, the callback is invoked with the local communications socket
644	as argument. At this point you can start using the socket in any way you	699	as argument. At this point you can start using the socket in any way you
645	like.	700	like.
646
647	The process object becomes unusable on return from this function - any
648	further method calls result in undefined behaviour.
649		701
650	If the communication socket isn't used, it should be closed on both sides,	702	If the communication socket isn't used, it should be closed on both sides,
651	to save on kernel memory.	703	to save on kernel memory.
652		704
653	The socket is non-blocking in the parent, and blocking in the newly	705	The socket is non-blocking in the parent, and blocking in the newly
…		…
728	479 vfork+execs per second, using AnyEvent::Fork->new_exec	780	479 vfork+execs per second, using AnyEvent::Fork->new_exec
729		781
730	So how can C<< AnyEvent->new >> be faster than a standard fork, even	782	So how can C<< AnyEvent->new >> be faster than a standard fork, even
731	though it uses the same operations, but adds a lot of overhead?	783	though it uses the same operations, but adds a lot of overhead?
732		784
733	The difference is simply the process size: forking the 6MB process takes	785	The difference is simply the process size: forking the 5MB process takes
734	so much longer than forking the 2.5MB template process that the overhead	786	so much longer than forking the 2.5MB template process that the extra
735	introduced is canceled out.	787	overhead introduced is canceled out.
736		788
737	If the benchmark process grows, the normal fork becomes even slower:	789	If the benchmark process grows, the normal fork becomes even slower:
738		790
739	1340 new processes, manual fork in a 20MB process	791	1340 new processes, manual fork of a 20MB process
740	731 new processes, manual fork in a 200MB process	792	731 new processes, manual fork of a 200MB process
741	235 new processes, manual fork in a 2000MB process	793	235 new processes, manual fork of a 2000MB process
742		794
743	What that means (to me) is that I can use this module without having a	795	What that means (to me) is that I can use this module without having a bad
744	very bad conscience because of the extra overhead required to start new	796	conscience because of the extra overhead required to start new processes.
745	processes.
746		797
747	=head1 TYPICAL PROBLEMS	798	=head1 TYPICAL PROBLEMS
748		799
749	This section lists typical problems that remain. I hope by recognising	800	This section lists typical problems that remain. I hope by recognising
750	them, most can be avoided.	801	them, most can be avoided.
751		802
752	=over 4	803	=over 4
753		804
754	=item "leaked" file descriptors for exec'ed processes	805	=item leaked file descriptors for exec'ed processes
755		806
756	POSIX systems inherit file descriptors by default when exec'ing a new	807	POSIX systems inherit file descriptors by default when exec'ing a new
757	process. While perl itself laudably sets the close-on-exec flags on new	808	process. While perl itself laudably sets the close-on-exec flags on new
758	file handles, most C libraries don't care, and even if all cared, it's	809	file handles, most C libraries don't care, and even if all cared, it's
759	often not possible to set the flag in a race-free manner.	810	often not possible to set the flag in a race-free manner.
…		…
779	libraries or the code that leaks those file descriptors.	830	libraries or the code that leaks those file descriptors.
780		831
781	Fortunately, most of these leaked descriptors do no harm, other than	832	Fortunately, most of these leaked descriptors do no harm, other than
782	sitting on some resources.	833	sitting on some resources.
783		834
784	=item "leaked" file descriptors for fork'ed processes	835	=item leaked file descriptors for fork'ed processes
785		836
786	Normally, L<AnyEvent::Fork> does start new processes by exec'ing them,	837	Normally, L<AnyEvent::Fork> does start new processes by exec'ing them,
787	which closes file descriptors not marked for being inherited.	838	which closes file descriptors not marked for being inherited.
788		839
789	However, L<AnyEvent::Fork::Early> and L<AnyEvent::Fork::Template> offer	840	However, L<AnyEvent::Fork::Early> and L<AnyEvent::Fork::Template> offer
…		…
798		849
799	The solution is to either not load these modules before use'ing	850	The solution is to either not load these modules before use'ing
800	L<AnyEvent::Fork::Early> or L<AnyEvent::Fork::Template>, or to delay	851	L<AnyEvent::Fork::Early> or L<AnyEvent::Fork::Template>, or to delay
801	initialising them, for example, by calling C<init Gtk2> manually.	852	initialising them, for example, by calling C<init Gtk2> manually.
802		853
803	=item exit runs destructors	854	=item exiting calls object destructors
804		855
805	This only applies to users of Lc<AnyEvent::Fork:Early> and	856	This only applies to users of Lc<AnyEvent::Fork:Early> and
806	L<AnyEvent::Fork::Template>.	857	L<AnyEvent::Fork::Template>.
807		858
808	When a process created by AnyEvent::Fork exits, it might do so by calling	859	When a process created by AnyEvent::Fork exits, it might do so by calling
…		…
830	to make it so, mostly due to the bloody broken perl that nobody seems to	881	to make it so, mostly due to the bloody broken perl that nobody seems to
831	care about. The fork emulation is a bad joke - I have yet to see something	882	care about. The fork emulation is a bad joke - I have yet to see something
832	useful that you can do with it without running into memory corruption	883	useful that you can do with it without running into memory corruption
833	issues or other braindamage. Hrrrr.	884	issues or other braindamage. Hrrrr.
834		885
835	Cygwin perl is not supported at the moment, as it should implement fd	886	Cygwin perl is not supported at the moment due to some hilarious
836	passing, but doesn't, and rolling my own is hard, as cygwin doesn't	887	shortcomings of its API - see L<IO::FDPoll> for more details.
837	support enough functionality to do it.
838		888
839	=head1 SEE ALSO	889	=head1 SEE ALSO
840		890
841	L<AnyEvent::Fork::Early> (to avoid executing a perl interpreter),	891	L<AnyEvent::Fork::Early> (to avoid executing a perl interpreter),
842	L<AnyEvent::Fork::Template> (to create a process by forking the main	892	L<AnyEvent::Fork::Template> (to create a process by forking the main

Diff Legend

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

Comparing AnyEvent-Fork/Fork.pm (file contents): Revision 1.24 by root, Sat Apr 6 08:32:23 2013 UTC vs. Revision 1.37 by root, Sat Apr 6 20:06:39 2013 UTC

Diff Legend

Comparing AnyEvent-Fork/Fork.pm (file contents):
Revision 1.24 by root, Sat Apr 6 08:32:23 2013 UTC vs.
Revision 1.37 by root, Sat Apr 6 20:06:39 2013 UTC