ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent-Fork/README
(Generate patch)

Comparing AnyEvent-Fork/README (file contents):
Revision 1.2 by root, Thu Apr 4 07:27:09 2013 UTC vs.
Revision 1.8 by root, Sun Apr 28 13:47:52 2013 UTC

1NAME 1NAME
2 AnyEvent::Fork - everything you wanted to use fork() for, but couldn't 2 AnyEvent::Fork - everything you wanted to use fork() for, but couldn't
3 3
4 ATTENTION, this is a very early release, and very untested. Consider it
5 a technology preview.
6
7SYNOPSIS 4SYNOPSIS
8 use AnyEvent::Fork; 5 use AnyEvent::Fork;
9 6
10 ################################################################## 7 AnyEvent::Fork
8 ->new
9 ->require ("MyModule")
10 ->run ("MyModule::server", my $cv = AE::cv);
11
12 my $fh = $cv->recv;
13
14DESCRIPTION
15 This module allows you to create new processes, without actually forking
16 them from your current process (avoiding the problems of forking), but
17 preserving most of the advantages of fork.
18
19 It can be used to create new worker processes or new independent
20 subprocesses for short- and long-running jobs, process pools (e.g. for
21 use in pre-forked servers) but also to spawn new external processes
22 (such as CGI scripts from a web server), which can be faster (and more
23 well behaved) than using fork+exec in big processes.
24
25 Special care has been taken to make this module useful from other
26 modules, while still supporting specialised environments such as
27 App::Staticperl or PAR::Packer.
28
29 WHAT THIS MODULE IS NOT
30 This module only creates processes and lets you pass file handles and
31 strings to it, and run perl code. It does not implement any kind of RPC
32 - there is no back channel from the process back to you, and there is no
33 RPC or message passing going on.
34
35 If you need some form of RPC, you could use the AnyEvent::Fork::RPC
36 companion module, which adds simple RPC/job queueing to a process
37 created by this module.
38
39 And if you need some automatic process pool management on top of
40 AnyEvent::Fork::RPC, you can look at the AnyEvent::Fork::Pool companion
41 module.
42
43 Or you can implement it yourself in whatever way you like: use some
44 message-passing module such as AnyEvent::MP, some pipe such as
45 AnyEvent::ZeroMQ, use AnyEvent::Handle on both sides to send e.g. JSON
46 or Storable messages, and so on.
47
48 COMPARISON TO OTHER MODULES
49 There is an abundance of modules on CPAN that do "something fork", such
50 as Parallel::ForkManager, AnyEvent::ForkManager, AnyEvent::Worker or
51 AnyEvent::Subprocess. There are modules that implement their own process
52 management, such as AnyEvent::DBI.
53
54 The problems that all these modules try to solve are real, however, none
55 of them (from what I have seen) tackle the very real problems of
56 unwanted memory sharing, efficiency, not being able to use event
57 processing or similar modules in the processes they create.
58
59 This module doesn't try to replace any of them - instead it tries to
60 solve the problem of creating processes with a minimum of fuss and
61 overhead (and also luxury). Ideally, most of these would use
62 AnyEvent::Fork internally, except they were written before AnyEvent:Fork
63 was available, so obviously had to roll their own.
64
65 PROBLEM STATEMENT
66 There are two traditional ways to implement parallel processing on UNIX
67 like operating systems - fork and process, and fork+exec and process.
68 They have different advantages and disadvantages that I describe below,
69 together with how this module tries to mitigate the disadvantages.
70
71 Forking from a big process can be very slow.
72 A 5GB process needs 0.05s to fork on my 3.6GHz amd64 GNU/Linux box.
73 This overhead is often shared with exec (because you have to fork
74 first), but in some circumstances (e.g. when vfork is used),
75 fork+exec can be much faster.
76
77 This module can help here by telling a small(er) helper process to
78 fork, which is faster then forking the main process, and also uses
79 vfork where possible. This gives the speed of vfork, with the
80 flexibility of fork.
81
82 Forking usually creates a copy-on-write copy of the parent process.
83 For example, modules or data files that are loaded will not use
84 additional memory after a fork. When exec'ing a new process, modules
85 and data files might need to be loaded again, at extra CPU and
86 memory cost. But when forking, literally all data structures are
87 copied - if the program frees them and replaces them by new data,
88 the child processes will retain the old version even if it isn't
89 used, which can suddenly and unexpectedly increase memory usage when
90 freeing memory.
91
92 The trade-off is between more sharing with fork (which can be good
93 or bad), and no sharing with exec.
94
95 This module allows the main program to do a controlled fork, and
96 allows modules to exec processes safely at any time. When creating a
97 custom process pool you can take advantage of data sharing via fork
98 without risking to share large dynamic data structures that will
99 blow up child memory usage.
100
101 In other words, this module puts you into control over what is being
102 shared and what isn't, at all times.
103
104 Exec'ing a new perl process might be difficult.
105 For example, it is not easy to find the correct path to the perl
106 interpreter - $^X might not be a perl interpreter at all.
107
108 This module tries hard to identify the correct path to the perl
109 interpreter. With a cooperative main program, exec'ing the
110 interpreter might not even be necessary, but even without help from
111 the main program, it will still work when used from a module.
112
113 Exec'ing a new perl process might be slow, as all necessary modules have
114 to be loaded from disk again, with no guarantees of success.
115 Long running processes might run into problems when perl is upgraded
116 and modules are no longer loadable because they refer to a different
117 perl version, or parts of a distribution are newer than the ones
118 already loaded.
119
120 This module supports creating pre-initialised perl processes to be
121 used as a template for new processes.
122
123 Forking might be impossible when a program is running.
124 For example, POSIX makes it almost impossible to fork from a
125 multi-threaded program while doing anything useful in the child - in
126 fact, if your perl program uses POSIX threads (even indirectly via
127 e.g. IO::AIO or threads), you cannot call fork on the perl level
128 anymore without risking corruption issues on a number of operating
129 systems.
130
131 This module can safely fork helper processes at any time, by calling
132 fork+exec in C, in a POSIX-compatible way (via Proc::FastSpawn).
133
134 Parallel processing with fork might be inconvenient or difficult to
135 implement. Modules might not work in both parent and child.
136 For example, when a program uses an event loop and creates watchers
137 it becomes very hard to use the event loop from a child program, as
138 the watchers already exist but are only meaningful in the parent.
139 Worse, a module might want to use such a module, not knowing whether
140 another module or the main program also does, leading to problems.
141
142 Apart from event loops, graphical toolkits also commonly fall into
143 the "unsafe module" category, or just about anything that
144 communicates with the external world, such as network libraries and
145 file I/O modules, which usually don't like being copied and then
146 allowed to continue in two processes.
147
148 With this module only the main program is allowed to create new
149 processes by forking (because only the main program can know when it
150 is still safe to do so) - all other processes are created via
151 fork+exec, which makes it possible to use modules such as event
152 loops or window interfaces safely.
153
154EXAMPLES
11 # create a single new process, tell it to run your worker function 155 Create a single new process, tell it to run your worker function.
12
13 AnyEvent::Fork 156 AnyEvent::Fork
14 ->new 157 ->new
15 ->require ("MyModule") 158 ->require ("MyModule")
16 ->run ("MyModule::worker, sub { 159 ->run ("MyModule::worker, sub {
17 my ($master_filehandle) = @_; 160 my ($master_filehandle) = @_;
18 161
19 # now $master_filehandle is connected to the 162 # now $master_filehandle is connected to the
20 # $slave_filehandle in the new process. 163 # $slave_filehandle in the new process.
21 }); 164 });
22 165
23 # MyModule::worker might look like this 166 "MyModule" might look like this:
167
168 package MyModule;
169
24 sub MyModule::worker { 170 sub worker {
25 my ($slave_filehandle) = @_; 171 my ($slave_filehandle) = @_;
26 172
27 # now $slave_filehandle is connected to the $master_filehandle 173 # now $slave_filehandle is connected to the $master_filehandle
28 # in the original prorcess. have fun! 174 # in the original prorcess. have fun!
29 } 175 }
30 176
31 ##################################################################
32 # create a pool of server processes all accepting on the same socket 177 Create a pool of server processes all accepting on the same socket.
33
34 # create listener socket 178 # create listener socket
35 my $listener = ...; 179 my $listener = ...;
36 180
37 # create a pool template, initialise it and give it the socket 181 # create a pool template, initialise it and give it the socket
38 my $pool = AnyEvent::Fork 182 my $pool = AnyEvent::Fork
… …
49 } 193 }
50 194
51 # now do other things - maybe use the filehandle provided by run 195 # now do other things - maybe use the filehandle provided by run
52 # to wait for the processes to die. or whatever. 196 # to wait for the processes to die. or whatever.
53 197
54 # My::Server::run might look like this 198 "My::Server" might look like this:
55 sub My::Server::run { 199
200 package My::Server;
201
202 sub run {
56 my ($slave, $listener, $id) = @_; 203 my ($slave, $listener, $id) = @_;
57 204
58 close $slave; # we do not use the socket, so close it to save resources 205 close $slave; # we do not use the socket, so close it to save resources
59 206
60 # we could go ballistic and use e.g. AnyEvent here, or IO::AIO, 207 # we could go ballistic and use e.g. AnyEvent here, or IO::AIO,
… …
62 while (my $socket = $listener->accept) { 209 while (my $socket = $listener->accept) {
63 # do sth. with new socket 210 # do sth. with new socket
64 } 211 }
65 } 212 }
66 213
67DESCRIPTION 214 use AnyEvent::Fork as a faster fork+exec
68 This module allows you to create new processes, without actually forking 215 This runs "/bin/echo hi", with standard output redirected to /tmp/log
69 them from your current process (avoiding the problems of forking), but 216 and standard error redirected to the communications socket. It is
70 preserving most of the advantages of fork. 217 usually faster than fork+exec, but still lets you prepare the
218 environment.
71 219
72 It can be used to create new worker processes or new independent 220 open my $output, ">/tmp/log" or die "$!";
73 subprocesses for short- and long-running jobs, process pools (e.g. for
74 use in pre-forked servers) but also to spawn new external processes
75 (such as CGI scripts from a webserver), which can be faster (and more
76 well behaved) than using fork+exec in big processes.
77 221
78 Special care has been taken to make this module useful from other 222 AnyEvent::Fork
79 modules, while still supporting specialised environments such as 223 ->new
224 ->eval ('
225 # compile a helper function for later use
226 sub run {
227 my ($fh, $output, @cmd) = @_;
228
229 # perl will clear close-on-exec on STDOUT/STDERR
230 open STDOUT, ">&", $output or die;
231 open STDERR, ">&", $fh or die;
232
233 exec @cmd;
234 }
235 ')
236 ->send_fh ($output)
237 ->send_arg ("/bin/echo", "hi")
238 ->run ("run", my $cv = AE::cv);
239
240 my $stderr = $cv->recv;
241
242 For stingy users: put the worker code into a "DATA" section.
243 When you want to be stingy with files, you cna put your code into the
244 "DATA" section of your module (or program):
245
246 use AnyEvent::Fork;
247
248 AnyEvent::Fork
249 ->new
250 ->eval (do { local $/; <DATA> })
251 ->run ("doit", sub { ... });
252
253 __DATA__
254
255 sub doit {
256 ... do something!
257 }
258
259 For stingy standalone programs: do not rely on external files at
260all.
261 For single-file scripts it can be inconvenient to rely on external files
262 - even when using < "DATA" section, you still need to "exec" an external
263 perl interpreter, which might not be available when using
80 App::Staticperl or PAR::Packer. 264 App::Staticperl, Urlader or PAR::Packer for example.
81 265
82PROBLEM STATEMENT 266 Two modules help here - AnyEvent::Fork::Early forks a template process
83 There are two ways to implement parallel processing on UNIX like 267 for all further calls to "new_exec", and AnyEvent::Fork::Template forks
84 operating systems - fork and process, and fork+exec and process. They 268 the main program as a template process.
85 have different advantages and disadvantages that I describe below,
86 together with how this module tries to mitigate the disadvantages.
87 269
88 Forking from a big process can be very slow (a 5GB process needs 0.05s 270 Here is how your main program should look like:
89 to fork on my 3.6GHz amd64 GNU/Linux box for example). This overhead is
90 often shared with exec (because you have to fork first), but in some
91 circumstances (e.g. when vfork is used), fork+exec can be much faster.
92 This module can help here by telling a small(er) helper process to
93 fork, or fork+exec instead.
94 271
95 Forking usually creates a copy-on-write copy of the parent process. 272 #! perl
96 Memory (for example, modules or data files that have been will not take
97 additional memory). When exec'ing a new process, modules and data files
98 might need to be loaded again, at extra cpu and memory cost. Likewise
99 when forking, all data structures are copied as well - if the program
100 frees them and replaces them by new data, the child processes will
101 retain the memory even if it isn't used.
102 This module allows the main program to do a controlled fork, and
103 allows modules to exec processes safely at any time. When creating a
104 custom process pool you can take advantage of data sharing via fork
105 without risking to share large dynamic data structures that will
106 blow up child memory usage.
107 273
108 Exec'ing a new perl process might be difficult and slow. For example, it 274 # optional, as the very first thing.
109 is not easy to find the correct path to the perl interpreter, and all 275 # in case modules want to create their own processes.
110 modules have to be loaded from disk again. Long running processes might 276 use AnyEvent::Fork::Early;
111 run into problems when perl is upgraded for example.
112 This module supports creating pre-initialised perl processes to be
113 used as template, and also tries hard to identify the correct path
114 to the perl interpreter. With a cooperative main program, exec'ing
115 the interpreter might not even be necessary.
116 277
117 Forking might be impossible when a program is running. For example, 278 # next, load all modules you need in your template process
118 POSIX makes it almost impossible to fork from a multithreaded program 279 use Example::My::Module
119 and do anything useful in the child - strictly speaking, if your perl 280 use Example::Whatever;
120 program uses posix threads (even indirectly via e.g. IO::AIO or
121 threads), you cannot call fork on the perl level anymore, at all.
122 This module can safely fork helper processes at any time, by caling
123 fork+exec in C, in a POSIX-compatible way.
124 281
125 Parallel processing with fork might be inconvenient or difficult to 282 # next, put your run function definition and anything else you
126 implement. For example, when a program uses an event loop and creates 283 # need, but do not use code outside of BEGIN blocks.
127 watchers it becomes very hard to use the event loop from a child 284 sub worker_run {
128 program, as the watchers already exist but are only meaningful in the 285 my ($fh, @args) = @_;
129 parent. Worse, a module might want to use such a system, not knowing 286 ...
130 whether another module or the main program also does, leading to 287 }
131 problems. 288
132 This module only lets the main program create pools by forking 289 # now preserve everything so far as AnyEvent::Fork object
133 (because only the main program can know when it is still safe to do 290 # in §TEMPLATE.
134 so) - all other pools are created by fork+exec, after which such 291 use AnyEvent::Fork::Template;
135 modules can again be loaded. 292
293 # do not put code outside of BEGIN blocks until here
294
295 # now use the $TEMPLATE process in any way you like
296
297 # for example: create 10 worker processes
298 my @worker;
299 my $cv = AE::cv;
300 for (1..10) {
301 $cv->begin;
302 $TEMPLATE->fork->send_arg ($_)->run ("worker_run", sub {
303 push @worker, shift;
304 $cv->end;
305 });
306 }
307 $cv->recv;
136 308
137CONCEPTS 309CONCEPTS
138 This module can create new processes either by executing a new perl 310 This module can create new processes either by executing a new perl
139 process, or by forking from an existing "template" process. 311 process, or by forking from an existing "template" process.
312
313 All these processes are called "child processes" (whether they are
314 direct children or not), while the process that manages them is called
315 the "parent process".
140 316
141 Each such process comes with its own file handle that can be used to 317 Each such process comes with its own file handle that can be used to
142 communicate with it (it's actually a socket - one end in the new 318 communicate with it (it's actually a socket - one end in the new
143 process, one end in the main process), and among the things you can do 319 process, one end in the main process), and among the things you can do
144 in it are load modules, fork new processes, send file handles to it, and 320 in it are load modules, fork new processes, send file handles to it, and
… …
154 memory used for the perl interpreter with the new process, but 330 memory used for the perl interpreter with the new process, but
155 loading modules takes time, and the memory is not shared with 331 loading modules takes time, and the memory is not shared with
156 anything else. 332 anything else.
157 333
158 This is ideal for when you only need one extra process of a kind, 334 This is ideal for when you only need one extra process of a kind,
159 with the option of starting and stipping it on demand. 335 with the option of starting and stopping it on demand.
160 336
161 Example: 337 Example:
162 338
163 AnyEvent::Fork 339 AnyEvent::Fork
164 ->new 340 ->new
… …
178 the modules you loaded) is shared between the processes, and each 354 the modules you loaded) is shared between the processes, and each
179 new process consumes relatively little memory of its own. 355 new process consumes relatively little memory of its own.
180 356
181 The disadvantage of this approach is that you need to create a 357 The disadvantage of this approach is that you need to create a
182 template process for the sole purpose of forking new processes from 358 template process for the sole purpose of forking new processes from
183 it, but if you only need a fixed number of proceses you can create 359 it, but if you only need a fixed number of processes you can create
184 them, and then destroy the template process. 360 them, and then destroy the template process.
185 361
186 Example: 362 Example:
187 363
188 my $template = AnyEvent::Fork->new->require ("Some::Module"); 364 my $template = AnyEvent::Fork->new->require ("Some::Module");
… …
212 ->require ("Some::Module") 388 ->require ("Some::Module")
213 ->run ("Some::Module::run", sub { 389 ->run ("Some::Module::run", sub {
214 my ($fork_fh) = @_; 390 my ($fork_fh) = @_;
215 }); 391 });
216 392
217FUNCTIONS 393THE "AnyEvent::Fork" CLASS
218 my $pool = new AnyEvent::Fork key => value... 394 This module exports nothing, and only implements a single class -
219 Create a new process pool. The following named parameters are 395 "AnyEvent::Fork".
220 supported: 396
397 There are two class constructors that both create new processes - "new"
398 and "new_exec". The "fork" method creates a new process by forking an
399 existing one and could be considered a third constructor.
400
401 Most of the remaining methods deal with preparing the new process, by
402 loading code, evaluating code and sending data to the new process. They
403 usually return the process object, so you can chain method calls.
404
405 If a process object is destroyed before calling its "run" method, then
406 the process simply exits. After "run" is called, all responsibility is
407 passed to the specified function.
408
409 As long as there is any outstanding work to be done, process objects
410 resist being destroyed, so there is no reason to store them unless you
411 need them later - configure and forget works just fine.
221 412
222 my $proc = new AnyEvent::Fork 413 my $proc = new AnyEvent::Fork
223 Create a new "empty" perl interpreter process and returns its 414 Create a new "empty" perl interpreter process and returns its
224 process object for further manipulation. 415 process object for further manipulation.
225 416
226 The new process is forked from a template process that is kept 417 The new process is forked from a template process that is kept
227 around for this purpose. When it doesn't exist yet, it is created by 418 around for this purpose. When it doesn't exist yet, it is created by
228 a call to "new_exec" and kept around for future calls. 419 a call to "new_exec" first and then stays around for future calls.
229
230 When the process object is destroyed, it will release the file
231 handle that connects it with the new process. When the new process
232 has not yet called "run", then the process will exit. Otherwise,
233 what happens depends entirely on the code that is executed.
234 420
235 $new_proc = $proc->fork 421 $new_proc = $proc->fork
236 Forks $proc, creating a new process, and returns the process object 422 Forks $proc, creating a new process, and returns the process object
237 of the new process. 423 of the new process.
238 424
… …
251 possible, and is also slower. 437 possible, and is also slower.
252 438
253 You should use "new" whenever possible, except when having a 439 You should use "new" whenever possible, except when having a
254 template process around is unacceptable. 440 template process around is unacceptable.
255 441
256 The path to the perl interpreter is divined usign various methods - 442 The path to the perl interpreter is divined using various methods -
257 first $^X is investigated to see if the path ends with something 443 first $^X is investigated to see if the path ends with something
258 that sounds as if it were the perl interpreter. Failing this, the 444 that sounds as if it were the perl interpreter. Failing this, the
259 module falls back to using $Config::Config{perlpath}. 445 module falls back to using $Config::Config{perlpath}.
260 446
447 The path to perl can also be overriden by setting the global
448 variable $AnyEvent::Fork::PERL - it's value will be used for all
449 subsequent invocations.
450
451 $pid = $proc->pid
452 Returns the process id of the process *iff it is a direct child of
453 the process running AnyEvent::Fork*, and "undef" otherwise.
454
455 Normally, only processes created via "AnyEvent::Fork->new_exec" and
456 AnyEvent::Fork::Template are direct children, and you are
457 responsible to clean up their zombies when they die.
458
459 All other processes are not direct children, and will be cleaned up
460 by AnyEvent::Fork itself.
461
261 $proc = $proc->eval ($perlcode, @args) 462 $proc = $proc->eval ($perlcode, @args)
262 Evaluates the given $perlcode as ... perl code, while setting @_ to 463 Evaluates the given $perlcode as ... Perl code, while setting @_ to
263 the strings specified by @args. 464 the strings specified by @args, in the "main" package.
264 465
265 This call is meant to do any custom initialisation that might be 466 This call is meant to do any custom initialisation that might be
266 required (for example, the "require" method uses it). It's not 467 required (for example, the "require" method uses it). It's not
267 supposed to be used to completely take over the process, use "run" 468 supposed to be used to completely take over the process, use "run"
268 for that. 469 for that.
… …
270 The code will usually be executed after this call returns, and there 471 The code will usually be executed after this call returns, and there
271 is no way to pass anything back to the calling process. Any 472 is no way to pass anything back to the calling process. Any
272 evaluation errors will be reported to stderr and cause the process 473 evaluation errors will be reported to stderr and cause the process
273 to exit. 474 to exit.
274 475
476 If you want to execute some code (that isn't in a module) to take
477 over the process, you should compile a function via "eval" first,
478 and then call it via "run". This also gives you access to any
479 arguments passed via the "send_xxx" methods, such as file handles.
480 See the "use AnyEvent::Fork as a faster fork+exec" example to see it
481 in action.
482
275 Returns the process object for easy chaining of method calls. 483 Returns the process object for easy chaining of method calls.
276 484
277 $proc = $proc->require ($module, ...) 485 $proc = $proc->require ($module, ...)
278 Tries to load the given module(s) into the process 486 Tries to load the given module(s) into the process
279 487
… …
281 489
282 $proc = $proc->send_fh ($handle, ...) 490 $proc = $proc->send_fh ($handle, ...)
283 Send one or more file handles (*not* file descriptors) to the 491 Send one or more file handles (*not* file descriptors) to the
284 process, to prepare a call to "run". 492 process, to prepare a call to "run".
285 493
286 The process object keeps a reference to the handles until this is 494 The process object keeps a reference to the handles until they have
287 done, so you must not explicitly close the handles. This is most 495 been passed over to the process, so you must not explicitly close
288 easily accomplished by simply not storing the file handles anywhere 496 the handles. This is most easily accomplished by simply not storing
289 after passing them to this method. 497 the file handles anywhere after passing them to this method - when
498 AnyEvent::Fork is finished using them, perl will automatically close
499 them.
290 500
291 Returns the process object for easy chaining of method calls. 501 Returns the process object for easy chaining of method calls.
292 502
293 Example: pass an fh to a process, and release it without closing. it 503 Example: pass a file handle to a process, and release it without
294 will be closed automatically when it is no longer used. 504 closing. It will be closed automatically when it is no longer used.
295 505
296 $proc->send_fh ($my_fh); 506 $proc->send_fh ($my_fh);
297 undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT 507 undef $my_fh; # free the reference if you want, but DO NOT CLOSE IT
298 508
299 $proc = $proc->send_arg ($string, ...) 509 $proc = $proc->send_arg ($string, ...)
300 Send one or more argument strings to the process, to prepare a call 510 Send one or more argument strings to the process, to prepare a call
301 to "run". The strings can be any octet string. 511 to "run". The strings can be any octet strings.
302 512
513 The protocol is optimised to pass a moderate number of relatively
514 short strings - while you can pass up to 4GB of data in one go, this
515 is more meant to pass some ID information or other startup info, not
516 big chunks of data.
517
303 Returns the process object for easy chaining of emthod calls. 518 Returns the process object for easy chaining of method calls.
304 519
305 $proc->run ($func, $cb->($fh)) 520 $proc->run ($func, $cb->($fh))
306 Enter the function specified by the fully qualified name in $func in 521 Enter the function specified by the function name in $func in the
307 the process. The function is called with the communication socket as 522 process. The function is called with the communication socket as
308 first argument, followed by all file handles and string arguments 523 first argument, followed by all file handles and string arguments
309 sent earlier via "send_fh" and "send_arg" methods, in the order they 524 sent earlier via "send_fh" and "send_arg" methods, in the order they
310 were called. 525 were called.
311 526
312 If the called function returns, the process exits.
313
314 Preparing the process can take time - when the process is ready, the
315 callback is invoked with the local communications socket as
316 argument.
317
318 The process object becomes unusable on return from this function. 527 The process object becomes unusable on return from this function -
528 any further method calls result in undefined behaviour.
529
530 The function name should be fully qualified, but if it isn't, it
531 will be looked up in the "main" package.
532
533 If the called function returns, doesn't exist, or any error occurs,
534 the process exits.
535
536 Preparing the process is done in the background - when all commands
537 have been sent, the callback is invoked with the local
538 communications socket as argument. At this point you can start using
539 the socket in any way you like.
319 540
320 If the communication socket isn't used, it should be closed on both 541 If the communication socket isn't used, it should be closed on both
321 sides, to save on kernel memory. 542 sides, to save on kernel memory.
322 543
323 The socket is non-blocking in the parent, and blocking in the newly 544 The socket is non-blocking in the parent, and blocking in the newly
324 created process. The close-on-exec flag is set on both. Even if not 545 created process. The close-on-exec flag is set in both.
546
325 used otherwise, the socket can be a good indicator for the existance 547 Even if not used otherwise, the socket can be a good indicator for
326 of the process - if the other process exits, you get a readable 548 the existence of the process - if the other process exits, you get a
327 event on it, because exiting the process closes the socket (if it 549 readable event on it, because exiting the process closes the socket
328 didn't create any children using fork). 550 (if it didn't create any children using fork).
329 551
330 Example: create a template for a process pool, pass a few strings, 552 Example: create a template for a process pool, pass a few strings,
331 some file handles, then fork, pass one more string, and run some 553 some file handles, then fork, pass one more string, and run some
332 code. 554 code.
333 555
… …
342 ->send_arg ("str3") 564 ->send_arg ("str3")
343 ->run ("Some::function", sub { 565 ->run ("Some::function", sub {
344 my ($fh) = @_; 566 my ($fh) = @_;
345 567
346 # fh is nonblocking, but we trust that the OS can accept these 568 # fh is nonblocking, but we trust that the OS can accept these
347 # extra 3 octets anyway. 569 # few octets anyway.
348 syswrite $fh, "hi #$_\n"; 570 syswrite $fh, "hi #$_\n";
349 571
350 # $fh is being closed here, as we don't store it anywhere 572 # $fh is being closed here, as we don't store it anywhere
351 }); 573 });
352 } 574 }
… …
354 # Some::function might look like this - all parameters passed before fork 576 # Some::function might look like this - all parameters passed before fork
355 # and after will be passed, in order, after the communications socket. 577 # and after will be passed, in order, after the communications socket.
356 sub Some::function { 578 sub Some::function {
357 my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_; 579 my ($fh, $str1, $str2, $fh1, $fh2, $str3) = @_;
358 580
359 print scalar <$fh>; # prints "hi 1\n" and "hi 2\n" 581 print scalar <$fh>; # prints "hi #1\n" and "hi #2\n" in any order
360 } 582 }
583
584 EXPERIMENTAL METHODS
585 These methods might go away completely or change behaviour, at any time.
586
587 $proc->to_fh ($cb->($fh)) # EXPERIMENTAL, MIGHT BE REMOVED
588 Flushes all commands out to the process and then calls the callback
589 with the communications socket.
590
591 The process object becomes unusable on return from this function -
592 any further method calls result in undefined behaviour.
593
594 The point of this method is to give you a file handle thta you cna
595 pass to another process. In that other process, you can call
596 "new_from_fh AnyEvent::Fork" to create a new "AnyEvent::Fork" object
597 from it, thereby effectively passing a fork object to another
598 process.
599
600 new_from_fh AnyEvent::Fork $fh # EXPERIMENTAL, MIGHT BE REMOVED
601 Takes a file handle originally rceeived by the "to_fh" method and
602 creates a new "AnyEvent:Fork" object. The child process itself will
603 not change in any way, i.e. it will keep all the modifications done
604 to it before calling "to_fh".
605
606 The new object is very much like the original object, except that
607 the "pid" method will return "undef" even if the process is a direct
608 child.
609
610PERFORMANCE
611 Now for some unscientific benchmark numbers (all done on an amd64
612 GNU/Linux box). These are intended to give you an idea of the relative
613 performance you can expect, they are not meant to be absolute
614 performance numbers.
615
616 OK, so, I ran a simple benchmark that creates a socket pair, forks,
617 calls exit in the child and waits for the socket to close in the parent.
618 I did load AnyEvent, EV and AnyEvent::Fork, for a total process size of
619 5100kB.
620
621 2079 new processes per second, using manual socketpair + fork
622
623 Then I did the same thing, but instead of calling fork, I called
624 AnyEvent::Fork->new->run ("CORE::exit") and then again waited for the
625 socket from the child to close on exit. This does the same thing as
626 manual socket pair + fork, except that what is forked is the template
627 process (2440kB), and the socket needs to be passed to the server at the
628 other end of the socket first.
629
630 2307 new processes per second, using AnyEvent::Fork->new
631
632 And finally, using "new_exec" instead "new", using vforks+execs to exec
633 a new perl interpreter and compile the small server each time, I get:
634
635 479 vfork+execs per second, using AnyEvent::Fork->new_exec
636
637 So how can "AnyEvent->new" be faster than a standard fork, even though
638 it uses the same operations, but adds a lot of overhead?
639
640 The difference is simply the process size: forking the 5MB process takes
641 so much longer than forking the 2.5MB template process that the extra
642 overhead is canceled out.
643
644 If the benchmark process grows, the normal fork becomes even slower:
645
646 1340 new processes, manual fork of a 20MB process
647 731 new processes, manual fork of a 200MB process
648 235 new processes, manual fork of a 2000MB process
649
650 What that means (to me) is that I can use this module without having a
651 bad conscience because of the extra overhead required to start new
652 processes.
653
654TYPICAL PROBLEMS
655 This section lists typical problems that remain. I hope by recognising
656 them, most can be avoided.
657
658 leaked file descriptors for exec'ed processes
659 POSIX systems inherit file descriptors by default when exec'ing a
660 new process. While perl itself laudably sets the close-on-exec flags
661 on new file handles, most C libraries don't care, and even if all
662 cared, it's often not possible to set the flag in a race-free
663 manner.
664
665 That means some file descriptors can leak through. And since it
666 isn't possible to know which file descriptors are "good" and
667 "necessary" (or even to know which file descriptors are open), there
668 is no good way to close the ones that might harm.
669
670 As an example of what "harm" can be done consider a web server that
671 accepts connections and afterwards some module uses AnyEvent::Fork
672 for the first time, causing it to fork and exec a new process, which
673 might inherit the network socket. When the server closes the socket,
674 it is still open in the child (which doesn't even know that) and the
675 client might conclude that the connection is still fine.
676
677 For the main program, there are multiple remedies available -
678 AnyEvent::Fork::Early is one, creating a process early and not using
679 "new_exec" is another, as in both cases, the first process can be
680 exec'ed well before many random file descriptors are open.
681
682 In general, the solution for these kind of problems is to fix the
683 libraries or the code that leaks those file descriptors.
684
685 Fortunately, most of these leaked descriptors do no harm, other than
686 sitting on some resources.
687
688 leaked file descriptors for fork'ed processes
689 Normally, AnyEvent::Fork does start new processes by exec'ing them,
690 which closes file descriptors not marked for being inherited.
691
692 However, AnyEvent::Fork::Early and AnyEvent::Fork::Template offer a
693 way to create these processes by forking, and this leaks more file
694 descriptors than exec'ing them, as there is no way to mark
695 descriptors as "close on fork".
696
697 An example would be modules like EV, IO::AIO or Gtk2. Both create
698 pipes for internal uses, and Gtk2 might open a connection to the X
699 server. EV and IO::AIO can deal with fork, but Gtk2 might have
700 trouble with a fork.
701
702 The solution is to either not load these modules before use'ing
703 AnyEvent::Fork::Early or AnyEvent::Fork::Template, or to delay
704 initialising them, for example, by calling "init Gtk2" manually.
705
706 exiting calls object destructors
707 This only applies to users of AnyEvent::Fork:Early and
708 AnyEvent::Fork::Template, or when initialising code creates objects
709 that reference external resources.
710
711 When a process created by AnyEvent::Fork exits, it might do so by
712 calling exit, or simply letting perl reach the end of the program.
713 At which point Perl runs all destructors.
714
715 Not all destructors are fork-safe - for example, an object that
716 represents the connection to an X display might tell the X server to
717 free resources, which is inconvenient when the "real" object in the
718 parent still needs to use them.
719
720 This is obviously not a problem for AnyEvent::Fork::Early, as you
721 used it as the very first thing, right?
722
723 It is a problem for AnyEvent::Fork::Template though - and the
724 solution is to not create objects with nontrivial destructors that
725 might have an effect outside of Perl.
361 726
362PORTABILITY NOTES 727PORTABILITY NOTES
363 Native win32 perls are somewhat supported (AnyEvent::Fork::Early is a 728 Native win32 perls are somewhat supported (AnyEvent::Fork::Early is a
364 nop, and ::Template is not going to work), and it cost a lot of blood 729 nop, and ::Template is not going to work), and it cost a lot of blood
365 and sweat to make it so, mostly due to the bloody broken perl that 730 and sweat to make it so, mostly due to the bloody broken perl that
366 nobody seems to care about. The fork emulation is a bad joke - I have 731 nobody seems to care about. The fork emulation is a bad joke - I have
367 yet to see something useful that you cna do with it without running into 732 yet to see something useful that you can do with it without running into
368 memory corruption issues or other braindamage. Hrrrr. 733 memory corruption issues or other braindamage. Hrrrr.
369 734
370 Cygwin perl is not supported at the moment, as it should implement fd 735 Since fork is endlessly broken on win32 perls (it doesn't even remotely
371 passing, but doesn't, and rolling my own is hard, as cygwin doesn't 736 work within it's documented limits) and quite obviously it's not getting
372 support enough functionality to do it. 737 improved any time soon, the best way to proceed on windows would be to
738 always use "new_exec" and thus never rely on perl's fork "emulation".
373 739
374AUTHOR 740 Cygwin perl is not supported at the moment due to some hilarious
741 shortcomings of its API - see IO::FDPoll for more details. If you never
742 use "send_fh" and always use "new_exec" to create processes, it should
743 work though.
744
745SEE ALSO
746 AnyEvent::Fork::Early, to avoid executing a perl interpreter at all
747 (part of this distribution).
748
749 AnyEvent::Fork::Template, to create a process by forking the main
750 program at a convenient time (part of this distribution).
751
752 AnyEvent::Fork::RPC, for simple RPC to child processes (on CPAN).
753
754 AnyEvent::Fork::Pool, for simple worker process pool (on CPAN).
755
756AUTHOR AND CONTACT INFORMATION
375 Marc Lehmann <schmorp@schmorp.de> 757 Marc Lehmann <schmorp@schmorp.de>
376 http://home.schmorp.de/ 758 http://software.schmorp.de/pkg/AnyEvent-Fork
377 759

Diff Legend

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