1 | =head1 NAME |
1 | =head1 NAME |
2 | |
2 | |
3 | AnyEvent::MP - multi-processing/message-passing framework |
3 | AnyEvent::MP - erlang-style multi-processing/message-passing framework |
4 | |
4 | |
5 | =head1 SYNOPSIS |
5 | =head1 SYNOPSIS |
6 | |
6 | |
7 | use AnyEvent::MP; |
7 | use AnyEvent::MP; |
8 | |
8 | |
… | |
… | |
30 | rcv $port, pong => sub { warn "pong received\n" }; |
30 | rcv $port, pong => sub { warn "pong received\n" }; |
31 | |
31 | |
32 | # create a port on another node |
32 | # create a port on another node |
33 | my $port = spawn $node, $initfunc, @initdata; |
33 | my $port = spawn $node, $initfunc, @initdata; |
34 | |
34 | |
|
|
35 | # destroy a port again |
|
|
36 | kil $port; # "normal" kill |
|
|
37 | kil $port, my_error => "everything is broken"; # error kill |
|
|
38 | |
35 | # monitoring |
39 | # monitoring |
36 | mon $port, $cb->(@msg) # callback is invoked on death |
40 | mon $port, $cb->(@msg) # callback is invoked on death |
37 | mon $port, $otherport # kill otherport on abnormal death |
41 | mon $port, $localport # kill localport on abnormal death |
38 | mon $port, $otherport, @msg # send message on death |
42 | mon $port, $localport, @msg # send message on death |
|
|
43 | |
|
|
44 | # temporarily execute code in port context |
|
|
45 | peval $port, sub { die "kill the port!" }; |
|
|
46 | |
|
|
47 | # execute callbacks in $SELF port context |
|
|
48 | my $timer = AE::timer 1, 0, psub { |
|
|
49 | die "kill the port, delayed"; |
|
|
50 | }; |
39 | |
51 | |
40 | =head1 CURRENT STATUS |
52 | =head1 CURRENT STATUS |
41 | |
53 | |
42 | bin/aemp - stable. |
54 | bin/aemp - stable. |
43 | AnyEvent::MP - stable API, should work. |
55 | AnyEvent::MP - stable API, should work. |
44 | AnyEvent::MP::Intro - explains most concepts. |
56 | AnyEvent::MP::Intro - explains most concepts. |
45 | AnyEvent::MP::Kernel - mostly stable. |
57 | AnyEvent::MP::Kernel - mostly stable API. |
46 | AnyEvent::MP::Global - stable but incomplete, protocol not yet final. |
58 | AnyEvent::MP::Global - stable API. |
47 | |
|
|
48 | stay tuned. |
|
|
49 | |
59 | |
50 | =head1 DESCRIPTION |
60 | =head1 DESCRIPTION |
51 | |
61 | |
52 | This module (-family) implements a simple message passing framework. |
62 | This module (-family) implements a simple message passing framework. |
53 | |
63 | |
… | |
… | |
68 | |
78 | |
69 | Ports allow you to register C<rcv> handlers that can match all or just |
79 | Ports allow you to register C<rcv> handlers that can match all or just |
70 | some messages. Messages send to ports will not be queued, regardless of |
80 | some messages. Messages send to ports will not be queued, regardless of |
71 | anything was listening for them or not. |
81 | anything was listening for them or not. |
72 | |
82 | |
|
|
83 | Ports are represented by (printable) strings called "port IDs". |
|
|
84 | |
73 | =item port ID - C<nodeid#portname> |
85 | =item port ID - C<nodeid#portname> |
74 | |
86 | |
75 | A port ID is the concatenation of a node ID, a hash-mark (C<#>) as |
87 | A port ID is the concatenation of a node ID, a hash-mark (C<#>) |
76 | separator, and a port name (a printable string of unspecified format). |
88 | as separator, and a port name (a printable string of unspecified |
|
|
89 | format created by AnyEvent::MP). |
77 | |
90 | |
78 | =item node |
91 | =item node |
79 | |
92 | |
80 | A node is a single process containing at least one port - the node port, |
93 | A node is a single process containing at least one port - the node port, |
81 | which enables nodes to manage each other remotely, and to create new |
94 | which enables nodes to manage each other remotely, and to create new |
82 | ports. |
95 | ports. |
83 | |
96 | |
84 | Nodes are either public (have one or more listening ports) or private |
97 | Nodes are either public (have one or more listening ports) or private |
85 | (no listening ports). Private nodes cannot talk to other private nodes |
98 | (no listening ports). Private nodes cannot talk to other private nodes |
86 | currently. |
99 | currently, but all nodes can talk to public nodes. |
87 | |
100 | |
|
|
101 | Nodes is represented by (printable) strings called "node IDs". |
|
|
102 | |
88 | =item node ID - C<[a-za-Z0-9_\-.:]+> |
103 | =item node ID - C<[A-Za-z0-9_\-.:]*> |
89 | |
104 | |
90 | A node ID is a string that uniquely identifies the node within a |
105 | A node ID is a string that uniquely identifies the node within a |
91 | network. Depending on the configuration used, node IDs can look like a |
106 | network. Depending on the configuration used, node IDs can look like a |
92 | hostname, a hostname and a port, or a random string. AnyEvent::MP itself |
107 | hostname, a hostname and a port, or a random string. AnyEvent::MP itself |
93 | doesn't interpret node IDs in any way. |
108 | doesn't interpret node IDs in any way except to uniquely identify a node. |
94 | |
109 | |
95 | =item binds - C<ip:port> |
110 | =item binds - C<ip:port> |
96 | |
111 | |
97 | Nodes can only talk to each other by creating some kind of connection to |
112 | Nodes can only talk to each other by creating some kind of connection to |
98 | each other. To do this, nodes should listen on one or more local transport |
113 | each other. To do this, nodes should listen on one or more local transport |
|
|
114 | endpoints - binds. |
|
|
115 | |
99 | endpoints - binds. Currently, only standard C<ip:port> specifications can |
116 | Currently, only standard C<ip:port> specifications can be used, which |
100 | be used, which specify TCP ports to listen on. |
117 | specify TCP ports to listen on. So a bind is basically just a tcp socket |
|
|
118 | in listening mode thta accepts conenctions form other nodes. |
101 | |
119 | |
|
|
120 | =item seed nodes |
|
|
121 | |
|
|
122 | When a node starts, it knows nothing about the network it is in - it |
|
|
123 | needs to connect to at least one other node that is already in the |
|
|
124 | network. These other nodes are called "seed nodes". |
|
|
125 | |
|
|
126 | Seed nodes themselves are not special - they are seed nodes only because |
|
|
127 | some other node I<uses> them as such, but any node can be used as seed |
|
|
128 | node for other nodes, and eahc node cna use a different set of seed nodes. |
|
|
129 | |
|
|
130 | In addition to discovering the network, seed nodes are also used to |
|
|
131 | maintain the network - all nodes using the same seed node form are part of |
|
|
132 | the same network. If a network is split into multiple subnets because e.g. |
|
|
133 | the network link between the parts goes down, then using the same seed |
|
|
134 | nodes for all nodes ensures that eventually the subnets get merged again. |
|
|
135 | |
|
|
136 | Seed nodes are expected to be long-running, and at least one seed node |
|
|
137 | should always be available. They should also be relatively responsive - a |
|
|
138 | seed node that blocks for long periods will slow down everybody else. |
|
|
139 | |
|
|
140 | For small networks, it's best if every node uses the same set of seed |
|
|
141 | nodes. For large networks, it can be useful to specify "regional" seed |
|
|
142 | nodes for most nodes in an area, and use all seed nodes as seed nodes for |
|
|
143 | each other. What's important is that all seed nodes connections form a |
|
|
144 | complete graph, so that the network cannot split into separate subnets |
|
|
145 | forever. |
|
|
146 | |
|
|
147 | Seed nodes are represented by seed IDs. |
|
|
148 | |
102 | =item seeds - C<host:port> |
149 | =item seed IDs - C<host:port> |
103 | |
150 | |
104 | When a node starts, it knows nothing about the network. To teach the node |
151 | Seed IDs are transport endpoint(s) (usually a hostname/IP address and a |
105 | about the network it first has to contact some other node within the |
152 | TCP port) of nodes that should be used as seed nodes. |
106 | network. This node is called a seed. |
|
|
107 | |
153 | |
108 | Seeds are transport endpoint(s) of as many nodes as one wants. Those nodes |
154 | =item global nodes |
109 | are expected to be long-running, and at least one of those should always |
|
|
110 | be available. When nodes run out of connections (e.g. due to a network |
|
|
111 | error), they try to re-establish connections to some seednodes again to |
|
|
112 | join the network. |
|
|
113 | |
155 | |
114 | Apart from being sued for seeding, seednodes are not special in any way - |
156 | An AEMP network needs a discovery service - nodes need to know how to |
115 | every public node can be a seednode. |
157 | connect to other nodes they only know by name. In addition, AEMP offers a |
|
|
158 | distributed "group database", which maps group names to a list of strings |
|
|
159 | - for example, to register worker ports. |
|
|
160 | |
|
|
161 | A network needs at least one global node to work, and allows every node to |
|
|
162 | be a global node. |
|
|
163 | |
|
|
164 | Any node that loads the L<AnyEvent::MP::Global> module becomes a global |
|
|
165 | node and tries to keep connections to all other nodes. So while it can |
|
|
166 | make sense to make every node "global" in small networks, it usually makes |
|
|
167 | sense to only make seed nodes into global nodes in large networks (nodes |
|
|
168 | keep connections to seed nodes and global nodes, so makign them the same |
|
|
169 | reduces overhead). |
116 | |
170 | |
117 | =back |
171 | =back |
118 | |
172 | |
119 | =head1 VARIABLES/FUNCTIONS |
173 | =head1 VARIABLES/FUNCTIONS |
120 | |
174 | |
… | |
… | |
122 | |
176 | |
123 | =cut |
177 | =cut |
124 | |
178 | |
125 | package AnyEvent::MP; |
179 | package AnyEvent::MP; |
126 | |
180 | |
|
|
181 | use AnyEvent::MP::Config (); |
127 | use AnyEvent::MP::Kernel; |
182 | use AnyEvent::MP::Kernel; |
|
|
183 | use AnyEvent::MP::Kernel qw(%NODE %PORT %PORT_DATA $UNIQ $RUNIQ $ID); |
128 | |
184 | |
129 | use common::sense; |
185 | use common::sense; |
130 | |
186 | |
131 | use Carp (); |
187 | use Carp (); |
132 | |
188 | |
133 | use AE (); |
189 | use AE (); |
|
|
190 | use Guard (); |
134 | |
191 | |
135 | use base "Exporter"; |
192 | use base "Exporter"; |
136 | |
193 | |
137 | our $VERSION = $AnyEvent::MP::Kernel::VERSION; |
194 | our $VERSION = $AnyEvent::MP::Config::VERSION; |
138 | |
195 | |
139 | our @EXPORT = qw( |
196 | our @EXPORT = qw( |
140 | NODE $NODE *SELF node_of after |
197 | NODE $NODE *SELF node_of after |
141 | configure |
198 | configure |
142 | snd rcv mon mon_guard kil reg psub spawn |
199 | snd rcv mon mon_guard kil psub peval spawn cal |
143 | port |
200 | port |
|
|
201 | db_set db_del db_reg |
|
|
202 | db_mon db_family db_keys db_values |
144 | ); |
203 | ); |
145 | |
204 | |
146 | our $SELF; |
205 | our $SELF; |
147 | |
206 | |
148 | sub _self_die() { |
207 | sub _self_die() { |
… | |
… | |
171 | some other nodes in the network to discover other nodes. |
230 | some other nodes in the network to discover other nodes. |
172 | |
231 | |
173 | This function configures a node - it must be called exactly once (or |
232 | This function configures a node - it must be called exactly once (or |
174 | never) before calling other AnyEvent::MP functions. |
233 | never) before calling other AnyEvent::MP functions. |
175 | |
234 | |
|
|
235 | The key/value pairs are basically the same ones as documented for the |
|
|
236 | F<aemp> command line utility (sans the set/del prefix), with these additions: |
|
|
237 | |
|
|
238 | =over 4 |
|
|
239 | |
|
|
240 | =item norc => $boolean (default false) |
|
|
241 | |
|
|
242 | If true, then the rc file (e.g. F<~/.perl-anyevent-mp>) will I<not> |
|
|
243 | be consulted - all configuraiton options must be specified in the |
|
|
244 | C<configure> call. |
|
|
245 | |
|
|
246 | =item force => $boolean (default false) |
|
|
247 | |
|
|
248 | IF true, then the values specified in the C<configure> will take |
|
|
249 | precedence over any values configured via the rc file. The default is for |
|
|
250 | the rc file to override any options specified in the program. |
|
|
251 | |
|
|
252 | =item secure => $pass->(@msg) |
|
|
253 | |
|
|
254 | In addition to specifying a boolean, you can specify a code reference that |
|
|
255 | is called for every code execution attempt - the execution request is |
|
|
256 | granted iff the callback returns a true value. |
|
|
257 | |
|
|
258 | Most of the time the callback should look only at |
|
|
259 | C<$AnyEvent::MP::Kernel::SRCNODE> to make a decision, and not at the |
|
|
260 | actual message (which can be about anything, and is mostly provided for |
|
|
261 | diagnostic purposes). |
|
|
262 | |
|
|
263 | See F<semp setsecure> for more info. |
|
|
264 | |
|
|
265 | =back |
|
|
266 | |
176 | =over 4 |
267 | =over 4 |
177 | |
268 | |
178 | =item step 1, gathering configuration from profiles |
269 | =item step 1, gathering configuration from profiles |
179 | |
270 | |
180 | The function first looks up a profile in the aemp configuration (see the |
271 | The function first looks up a profile in the aemp configuration (see the |
… | |
… | |
193 | That means that the values specified in the profile have highest priority |
284 | That means that the values specified in the profile have highest priority |
194 | and the values specified directly via C<configure> have lowest priority, |
285 | and the values specified directly via C<configure> have lowest priority, |
195 | and can only be used to specify defaults. |
286 | and can only be used to specify defaults. |
196 | |
287 | |
197 | If the profile specifies a node ID, then this will become the node ID of |
288 | If the profile specifies a node ID, then this will become the node ID of |
198 | this process. If not, then the profile name will be used as node ID. The |
289 | this process. If not, then the profile name will be used as node ID, with |
199 | special node ID of C<anon/> will be replaced by a random node ID. |
290 | a unique randoms tring (C</%u>) appended. |
|
|
291 | |
|
|
292 | The node ID can contain some C<%> sequences that are expanded: C<%n> |
|
|
293 | is expanded to the local nodename, C<%u> is replaced by a random |
|
|
294 | strign to make the node unique. For example, the F<aemp> commandline |
|
|
295 | utility uses C<aemp/%n/%u> as nodename, which might expand to |
|
|
296 | C<aemp/cerebro/ZQDGSIkRhEZQDGSIkRhE>. |
200 | |
297 | |
201 | =item step 2, bind listener sockets |
298 | =item step 2, bind listener sockets |
202 | |
299 | |
203 | The next step is to look up the binds in the profile, followed by binding |
300 | The next step is to look up the binds in the profile, followed by binding |
204 | aemp protocol listeners on all binds specified (it is possible and valid |
301 | aemp protocol listeners on all binds specified (it is possible and valid |
… | |
… | |
210 | used, meaning the node will bind on a dynamically-assigned port on every |
307 | used, meaning the node will bind on a dynamically-assigned port on every |
211 | local IP address it finds. |
308 | local IP address it finds. |
212 | |
309 | |
213 | =item step 3, connect to seed nodes |
310 | =item step 3, connect to seed nodes |
214 | |
311 | |
215 | As the last step, the seeds list from the profile is passed to the |
312 | As the last step, the seed ID list from the profile is passed to the |
216 | L<AnyEvent::MP::Global> module, which will then use it to keep |
313 | L<AnyEvent::MP::Global> module, which will then use it to keep |
217 | connectivity with at least one node at any point in time. |
314 | connectivity with at least one node at any point in time. |
218 | |
315 | |
219 | =back |
316 | =back |
220 | |
317 | |
221 | Example: become a distributed node using the locla node name as profile. |
318 | Example: become a distributed node using the local node name as profile. |
222 | This should be the most common form of invocation for "daemon"-type nodes. |
319 | This should be the most common form of invocation for "daemon"-type nodes. |
223 | |
320 | |
224 | configure |
321 | configure |
225 | |
322 | |
226 | Example: become an anonymous node. This form is often used for commandline |
323 | Example: become a semi-anonymous node. This form is often used for |
227 | clients. |
324 | commandline clients. |
228 | |
325 | |
229 | configure nodeid => "anon/"; |
326 | configure nodeid => "myscript/%n/%u"; |
230 | |
327 | |
231 | Example: configure a node using a profile called seed, which si suitable |
328 | Example: configure a node using a profile called seed, which is suitable |
232 | for a seed node as it binds on all local addresses on a fixed port (4040, |
329 | for a seed node as it binds on all local addresses on a fixed port (4040, |
233 | customary for aemp). |
330 | customary for aemp). |
234 | |
331 | |
235 | # use the aemp commandline utility |
332 | # use the aemp commandline utility |
236 | # aemp profile seed nodeid anon/ binds '*:4040' |
333 | # aemp profile seed binds '*:4040' |
237 | |
334 | |
238 | # then use it |
335 | # then use it |
239 | configure profile => "seed"; |
336 | configure profile => "seed"; |
240 | |
337 | |
241 | # or simply use aemp from the shell again: |
338 | # or simply use aemp from the shell again: |
… | |
… | |
306 | |
403 | |
307 | =cut |
404 | =cut |
308 | |
405 | |
309 | sub rcv($@); |
406 | sub rcv($@); |
310 | |
407 | |
311 | sub _kilme { |
408 | my $KILME = sub { |
312 | die "received message on port without callback"; |
409 | (my $tag = substr $_[0], 0, 30) =~ s/([\x20-\x7e])/./g; |
313 | } |
410 | kil $SELF, unhandled_message => "no callback found for message '$tag'"; |
|
|
411 | }; |
314 | |
412 | |
315 | sub port(;&) { |
413 | sub port(;&) { |
316 | my $id = "$UNIQ." . $ID++; |
414 | my $id = $UNIQ . ++$ID; |
317 | my $port = "$NODE#$id"; |
415 | my $port = "$NODE#$id"; |
318 | |
416 | |
319 | rcv $port, shift || \&_kilme; |
417 | rcv $port, shift || $KILME; |
320 | |
418 | |
321 | $port |
419 | $port |
322 | } |
420 | } |
323 | |
421 | |
324 | =item rcv $local_port, $callback->(@msg) |
422 | =item rcv $local_port, $callback->(@msg) |
… | |
… | |
329 | |
427 | |
330 | The global C<$SELF> (exported by this module) contains C<$port> while |
428 | The global C<$SELF> (exported by this module) contains C<$port> while |
331 | executing the callback. Runtime errors during callback execution will |
429 | executing the callback. Runtime errors during callback execution will |
332 | result in the port being C<kil>ed. |
430 | result in the port being C<kil>ed. |
333 | |
431 | |
334 | The default callback received all messages not matched by a more specific |
432 | The default callback receives all messages not matched by a more specific |
335 | C<tag> match. |
433 | C<tag> match. |
336 | |
434 | |
337 | =item rcv $local_port, tag => $callback->(@msg_without_tag), ... |
435 | =item rcv $local_port, tag => $callback->(@msg_without_tag), ... |
338 | |
436 | |
339 | Register (or replace) callbacks to be called on messages starting with the |
437 | Register (or replace) callbacks to be called on messages starting with the |
… | |
… | |
360 | msg1 => sub { ... }, |
458 | msg1 => sub { ... }, |
361 | ... |
459 | ... |
362 | ; |
460 | ; |
363 | |
461 | |
364 | Example: temporarily register a rcv callback for a tag matching some port |
462 | Example: temporarily register a rcv callback for a tag matching some port |
365 | (e.g. for a rpc reply) and unregister it after a message was received. |
463 | (e.g. for an rpc reply) and unregister it after a message was received. |
366 | |
464 | |
367 | rcv $port, $otherport => sub { |
465 | rcv $port, $otherport => sub { |
368 | my @reply = @_; |
466 | my @reply = @_; |
369 | |
467 | |
370 | rcv $SELF, $otherport; |
468 | rcv $SELF, $otherport; |
… | |
… | |
383 | if (ref $_[0]) { |
481 | if (ref $_[0]) { |
384 | if (my $self = $PORT_DATA{$portid}) { |
482 | if (my $self = $PORT_DATA{$portid}) { |
385 | "AnyEvent::MP::Port" eq ref $self |
483 | "AnyEvent::MP::Port" eq ref $self |
386 | or Carp::croak "$port: rcv can only be called on message matching ports, caught"; |
484 | or Carp::croak "$port: rcv can only be called on message matching ports, caught"; |
387 | |
485 | |
388 | $self->[2] = shift; |
486 | $self->[0] = shift; |
389 | } else { |
487 | } else { |
390 | my $cb = shift; |
488 | my $cb = shift; |
391 | $PORT{$portid} = sub { |
489 | $PORT{$portid} = sub { |
392 | local $SELF = $port; |
490 | local $SELF = $port; |
393 | eval { &$cb }; _self_die if $@; |
491 | eval { &$cb }; _self_die if $@; |
394 | }; |
492 | }; |
395 | } |
493 | } |
396 | } elsif (defined $_[0]) { |
494 | } elsif (defined $_[0]) { |
397 | my $self = $PORT_DATA{$portid} ||= do { |
495 | my $self = $PORT_DATA{$portid} ||= do { |
398 | my $self = bless [$PORT{$port} || sub { }, { }, $port], "AnyEvent::MP::Port"; |
496 | my $self = bless [$PORT{$portid} || sub { }, { }, $port], "AnyEvent::MP::Port"; |
399 | |
497 | |
400 | $PORT{$portid} = sub { |
498 | $PORT{$portid} = sub { |
401 | local $SELF = $port; |
499 | local $SELF = $port; |
402 | |
500 | |
403 | if (my $cb = $self->[1]{$_[0]}) { |
501 | if (my $cb = $self->[1]{$_[0]}) { |
… | |
… | |
425 | } |
523 | } |
426 | |
524 | |
427 | $port |
525 | $port |
428 | } |
526 | } |
429 | |
527 | |
|
|
528 | =item peval $port, $coderef[, @args] |
|
|
529 | |
|
|
530 | Evaluates the given C<$codref> within the contetx of C<$port>, that is, |
|
|
531 | when the code throews an exception the C<$port> will be killed. |
|
|
532 | |
|
|
533 | Any remaining args will be passed to the callback. Any return values will |
|
|
534 | be returned to the caller. |
|
|
535 | |
|
|
536 | This is useful when you temporarily want to execute code in the context of |
|
|
537 | a port. |
|
|
538 | |
|
|
539 | Example: create a port and run some initialisation code in it's context. |
|
|
540 | |
|
|
541 | my $port = port { ... }; |
|
|
542 | |
|
|
543 | peval $port, sub { |
|
|
544 | init |
|
|
545 | or die "unable to init"; |
|
|
546 | }; |
|
|
547 | |
|
|
548 | =cut |
|
|
549 | |
|
|
550 | sub peval($$) { |
|
|
551 | local $SELF = shift; |
|
|
552 | my $cb = shift; |
|
|
553 | |
|
|
554 | if (wantarray) { |
|
|
555 | my @res = eval { &$cb }; |
|
|
556 | _self_die if $@; |
|
|
557 | @res |
|
|
558 | } else { |
|
|
559 | my $res = eval { &$cb }; |
|
|
560 | _self_die if $@; |
|
|
561 | $res |
|
|
562 | } |
|
|
563 | } |
|
|
564 | |
430 | =item $closure = psub { BLOCK } |
565 | =item $closure = psub { BLOCK } |
431 | |
566 | |
432 | Remembers C<$SELF> and creates a closure out of the BLOCK. When the |
567 | Remembers C<$SELF> and creates a closure out of the BLOCK. When the |
433 | closure is executed, sets up the environment in the same way as in C<rcv> |
568 | closure is executed, sets up the environment in the same way as in C<rcv> |
434 | callbacks, i.e. runtime errors will cause the port to get C<kil>ed. |
569 | callbacks, i.e. runtime errors will cause the port to get C<kil>ed. |
|
|
570 | |
|
|
571 | The effect is basically as if it returned C<< sub { peval $SELF, sub { |
|
|
572 | BLOCK }, @_ } >>. |
435 | |
573 | |
436 | This is useful when you register callbacks from C<rcv> callbacks: |
574 | This is useful when you register callbacks from C<rcv> callbacks: |
437 | |
575 | |
438 | rcv delayed_reply => sub { |
576 | rcv delayed_reply => sub { |
439 | my ($delay, @reply) = @_; |
577 | my ($delay, @reply) = @_; |
… | |
… | |
512 | delivered again. |
650 | delivered again. |
513 | |
651 | |
514 | Inter-host-connection timeouts and monitoring depend on the transport |
652 | Inter-host-connection timeouts and monitoring depend on the transport |
515 | used. The only transport currently implemented is TCP, and AnyEvent::MP |
653 | used. The only transport currently implemented is TCP, and AnyEvent::MP |
516 | relies on TCP to detect node-downs (this can take 10-15 minutes on a |
654 | relies on TCP to detect node-downs (this can take 10-15 minutes on a |
517 | non-idle connection, and usually around two hours for idle conenctions). |
655 | non-idle connection, and usually around two hours for idle connections). |
518 | |
656 | |
519 | This means that monitoring is good for program errors and cleaning up |
657 | This means that monitoring is good for program errors and cleaning up |
520 | stuff eventually, but they are no replacement for a timeout when you need |
658 | stuff eventually, but they are no replacement for a timeout when you need |
521 | to ensure some maximum latency. |
659 | to ensure some maximum latency. |
522 | |
660 | |
… | |
… | |
554 | } |
692 | } |
555 | |
693 | |
556 | $node->monitor ($port, $cb); |
694 | $node->monitor ($port, $cb); |
557 | |
695 | |
558 | defined wantarray |
696 | defined wantarray |
559 | and AnyEvent::Util::guard { $node->unmonitor ($port, $cb) } |
697 | and ($cb += 0, Guard::guard { $node->unmonitor ($port, $cb) }) |
560 | } |
698 | } |
561 | |
699 | |
562 | =item $guard = mon_guard $port, $ref, $ref... |
700 | =item $guard = mon_guard $port, $ref, $ref... |
563 | |
701 | |
564 | Monitors the given C<$port> and keeps the passed references. When the port |
702 | Monitors the given C<$port> and keeps the passed references. When the port |
… | |
… | |
587 | |
725 | |
588 | =item kil $port[, @reason] |
726 | =item kil $port[, @reason] |
589 | |
727 | |
590 | Kill the specified port with the given C<@reason>. |
728 | Kill the specified port with the given C<@reason>. |
591 | |
729 | |
592 | If no C<@reason> is specified, then the port is killed "normally" (ports |
730 | If no C<@reason> is specified, then the port is killed "normally" - |
593 | monitoring other ports will not necessarily die because a port dies |
731 | monitor callback will be invoked, but the kil will not cause linked ports |
594 | "normally"). |
732 | (C<mon $mport, $lport> form) to get killed. |
595 | |
733 | |
596 | Otherwise, linked ports get killed with the same reason (second form of |
734 | If a C<@reason> is specified, then linked ports (C<mon $mport, $lport> |
597 | C<mon>, see above). |
735 | form) get killed with the same reason. |
598 | |
736 | |
599 | Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks |
737 | Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks |
600 | will be reported as reason C<< die => $@ >>. |
738 | will be reported as reason C<< die => $@ >>. |
601 | |
739 | |
602 | Transport/communication errors are reported as C<< transport_error => |
740 | Transport/communication errors are reported as C<< transport_error => |
603 | $message >>. |
741 | $message >>. |
604 | |
742 | |
605 | =cut |
743 | Common idioms: |
|
|
744 | |
|
|
745 | # silently remove yourself, do not kill linked ports |
|
|
746 | kil $SELF; |
|
|
747 | |
|
|
748 | # report a failure in some detail |
|
|
749 | kil $SELF, failure_mode_1 => "it failed with too high temperature"; |
|
|
750 | |
|
|
751 | # do not waste much time with killing, just die when something goes wrong |
|
|
752 | open my $fh, "<file" |
|
|
753 | or die "file: $!"; |
606 | |
754 | |
607 | =item $port = spawn $node, $initfunc[, @initdata] |
755 | =item $port = spawn $node, $initfunc[, @initdata] |
608 | |
756 | |
609 | Creates a port on the node C<$node> (which can also be a port ID, in which |
757 | Creates a port on the node C<$node> (which can also be a port ID, in which |
610 | case it's the node where that port resides). |
758 | case it's the node where that port resides). |
… | |
… | |
621 | the package, then the package above the package and so on (e.g. |
769 | the package, then the package above the package and so on (e.g. |
622 | C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function |
770 | C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function |
623 | exists or it runs out of package names. |
771 | exists or it runs out of package names. |
624 | |
772 | |
625 | The init function is then called with the newly-created port as context |
773 | The init function is then called with the newly-created port as context |
626 | object (C<$SELF>) and the C<@initdata> values as arguments. |
774 | object (C<$SELF>) and the C<@initdata> values as arguments. It I<must> |
|
|
775 | call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise |
|
|
776 | the port might not get created. |
627 | |
777 | |
628 | A common idiom is to pass a local port, immediately monitor the spawned |
778 | A common idiom is to pass a local port, immediately monitor the spawned |
629 | port, and in the remote init function, immediately monitor the passed |
779 | port, and in the remote init function, immediately monitor the passed |
630 | local port. This two-way monitoring ensures that both ports get cleaned up |
780 | local port. This two-way monitoring ensures that both ports get cleaned up |
631 | when there is a problem. |
781 | when there is a problem. |
… | |
… | |
655 | |
805 | |
656 | sub _spawn { |
806 | sub _spawn { |
657 | my $port = shift; |
807 | my $port = shift; |
658 | my $init = shift; |
808 | my $init = shift; |
659 | |
809 | |
|
|
810 | # rcv will create the actual port |
660 | local $SELF = "$NODE#$port"; |
811 | local $SELF = "$NODE#$port"; |
661 | eval { |
812 | eval { |
662 | &{ load_func $init } |
813 | &{ load_func $init } |
663 | }; |
814 | }; |
664 | _self_die if $@; |
815 | _self_die if $@; |
665 | } |
816 | } |
666 | |
817 | |
667 | sub spawn(@) { |
818 | sub spawn(@) { |
668 | my ($nodeid, undef) = split /#/, shift, 2; |
819 | my ($nodeid, undef) = split /#/, shift, 2; |
669 | |
820 | |
670 | my $id = "$RUNIQ." . $ID++; |
821 | my $id = $RUNIQ . ++$ID; |
671 | |
822 | |
672 | $_[0] =~ /::/ |
823 | $_[0] =~ /::/ |
673 | or Carp::croak "spawn init function must be a fully-qualified name, caught"; |
824 | or Carp::croak "spawn init function must be a fully-qualified name, caught"; |
674 | |
825 | |
675 | snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_; |
826 | snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_; |
676 | |
827 | |
677 | "$nodeid#$id" |
828 | "$nodeid#$id" |
678 | } |
829 | } |
|
|
830 | |
679 | |
831 | |
680 | =item after $timeout, @msg |
832 | =item after $timeout, @msg |
681 | |
833 | |
682 | =item after $timeout, $callback |
834 | =item after $timeout, $callback |
683 | |
835 | |
… | |
… | |
699 | ? $action[0]() |
851 | ? $action[0]() |
700 | : snd @action; |
852 | : snd @action; |
701 | }; |
853 | }; |
702 | } |
854 | } |
703 | |
855 | |
|
|
856 | #=item $cb2 = timeout $seconds, $cb[, @args] |
|
|
857 | |
|
|
858 | =item cal $port, @msg, $callback[, $timeout] |
|
|
859 | |
|
|
860 | A simple form of RPC - sends a message to the given C<$port> with the |
|
|
861 | given contents (C<@msg>), but adds a reply port to the message. |
|
|
862 | |
|
|
863 | The reply port is created temporarily just for the purpose of receiving |
|
|
864 | the reply, and will be C<kil>ed when no longer needed. |
|
|
865 | |
|
|
866 | A reply message sent to the port is passed to the C<$callback> as-is. |
|
|
867 | |
|
|
868 | If an optional time-out (in seconds) is given and it is not C<undef>, |
|
|
869 | then the callback will be called without any arguments after the time-out |
|
|
870 | elapsed and the port is C<kil>ed. |
|
|
871 | |
|
|
872 | If no time-out is given (or it is C<undef>), then the local port will |
|
|
873 | monitor the remote port instead, so it eventually gets cleaned-up. |
|
|
874 | |
|
|
875 | Currently this function returns the temporary port, but this "feature" |
|
|
876 | might go in future versions unless you can make a convincing case that |
|
|
877 | this is indeed useful for something. |
|
|
878 | |
|
|
879 | =cut |
|
|
880 | |
|
|
881 | sub cal(@) { |
|
|
882 | my $timeout = ref $_[-1] ? undef : pop; |
|
|
883 | my $cb = pop; |
|
|
884 | |
|
|
885 | my $port = port { |
|
|
886 | undef $timeout; |
|
|
887 | kil $SELF; |
|
|
888 | &$cb; |
|
|
889 | }; |
|
|
890 | |
|
|
891 | if (defined $timeout) { |
|
|
892 | $timeout = AE::timer $timeout, 0, sub { |
|
|
893 | undef $timeout; |
|
|
894 | kil $port; |
|
|
895 | $cb->(); |
|
|
896 | }; |
|
|
897 | } else { |
|
|
898 | mon $_[0], sub { |
|
|
899 | kil $port; |
|
|
900 | $cb->(); |
|
|
901 | }; |
|
|
902 | } |
|
|
903 | |
|
|
904 | push @_, $port; |
|
|
905 | &snd; |
|
|
906 | |
|
|
907 | $port |
|
|
908 | } |
|
|
909 | |
|
|
910 | =back |
|
|
911 | |
|
|
912 | =head1 DISTRIBUTED DATABASE |
|
|
913 | |
|
|
914 | AnyEvent::MP comes with a simple distributed database. The database will |
|
|
915 | be mirrored asynchronously on all global nodes. Other nodes bind to one |
|
|
916 | of the global nodes for their needs. Every node has a "local database" |
|
|
917 | which contains all the values that are set locally. All local databases |
|
|
918 | are merged together to form the global database, which can be queried. |
|
|
919 | |
|
|
920 | The database structure is that of a two-level hash - the database hash |
|
|
921 | contains hashes which contain values, similarly to a perl hash of hashes, |
|
|
922 | i.e.: |
|
|
923 | |
|
|
924 | $DATABASE{$family}{$subkey} = $value |
|
|
925 | |
|
|
926 | The top level hash key is called "family", and the second-level hash key |
|
|
927 | is called "subkey" or simply "key". |
|
|
928 | |
|
|
929 | The family must be alphanumeric, i.e. start with a letter and consist |
|
|
930 | of letters, digits, underscores and colons (C<[A-Za-z][A-Za-z0-9_:]*>, |
|
|
931 | pretty much like Perl module names. |
|
|
932 | |
|
|
933 | As the family namespace is global, it is recommended to prefix family names |
|
|
934 | with the name of the application or module using it. |
|
|
935 | |
|
|
936 | The subkeys must be non-empty strings, with no further restrictions. |
|
|
937 | |
|
|
938 | The values should preferably be strings, but other perl scalars should |
|
|
939 | work as well (such as C<undef>, arrays and hashes). |
|
|
940 | |
|
|
941 | Every database entry is owned by one node - adding the same family/subkey |
|
|
942 | combination on multiple nodes will not cause discomfort for AnyEvent::MP, |
|
|
943 | but the result might be nondeterministic, i.e. the key might have |
|
|
944 | different values on different nodes. |
|
|
945 | |
|
|
946 | Different subkeys in the same family can be owned by different nodes |
|
|
947 | without problems, and in fact, this is the common method to create worker |
|
|
948 | pools. For example, a worker port for image scaling might do this: |
|
|
949 | |
|
|
950 | db_set my_image_scalers => $port; |
|
|
951 | |
|
|
952 | And clients looking for an image scaler will want to get the |
|
|
953 | C<my_image_scalers> keys from time to time: |
|
|
954 | |
|
|
955 | db_keys my_image_scalers => sub { |
|
|
956 | @ports = @{ $_[0] }; |
|
|
957 | }; |
|
|
958 | |
|
|
959 | Or better yet, they want to monitor the database family, so they always |
|
|
960 | have a reasonable up-to-date copy: |
|
|
961 | |
|
|
962 | db_mon my_image_scalers => sub { |
|
|
963 | @ports = keys %{ $_[0] }; |
|
|
964 | }; |
|
|
965 | |
|
|
966 | In general, you can set or delete single subkeys, but query and monitor |
|
|
967 | whole families only. |
|
|
968 | |
|
|
969 | If you feel the need to monitor or query a single subkey, try giving it |
|
|
970 | it's own family. |
|
|
971 | |
|
|
972 | =over |
|
|
973 | |
|
|
974 | =item $guard = db_set $family => $subkey [=> $value] |
|
|
975 | |
|
|
976 | Sets (or replaces) a key to the database - if C<$value> is omitted, |
|
|
977 | C<undef> is used instead. |
|
|
978 | |
|
|
979 | When called in non-void context, C<db_set> returns a guard that |
|
|
980 | automatically calls C<db_del> when it is destroyed. |
|
|
981 | |
|
|
982 | =item db_del $family => $subkey... |
|
|
983 | |
|
|
984 | Deletes one or more subkeys from the database family. |
|
|
985 | |
|
|
986 | =item $guard = db_reg $family => $port => $value |
|
|
987 | |
|
|
988 | =item $guard = db_reg $family => $port |
|
|
989 | |
|
|
990 | =item $guard = db_reg $family |
|
|
991 | |
|
|
992 | Registers a port in the given family and optionally returns a guard to |
|
|
993 | remove it. |
|
|
994 | |
|
|
995 | This function basically does the same as: |
|
|
996 | |
|
|
997 | db_set $family => $port => $value |
|
|
998 | |
|
|
999 | Except that the port is monitored and automatically removed from the |
|
|
1000 | database family when it is kil'ed. |
|
|
1001 | |
|
|
1002 | If C<$value> is missing, C<undef> is used. If C<$port> is missing, then |
|
|
1003 | C<$SELF> is used. |
|
|
1004 | |
|
|
1005 | This function is most useful to register a port in some port group (which |
|
|
1006 | is just another name for a database family), and have it removed when the |
|
|
1007 | port is gone. This works best when the port is a local port. |
|
|
1008 | |
|
|
1009 | =cut |
|
|
1010 | |
|
|
1011 | sub db_reg($$;$) { |
|
|
1012 | my $family = shift; |
|
|
1013 | my $port = @_ ? shift : $SELF; |
|
|
1014 | |
|
|
1015 | my $clr = sub { db_del $family => $port }; |
|
|
1016 | mon $port, $clr; |
|
|
1017 | |
|
|
1018 | db_set $family => $port => $_[0]; |
|
|
1019 | |
|
|
1020 | defined wantarray |
|
|
1021 | and &Guard::guard ($clr) |
|
|
1022 | } |
|
|
1023 | |
|
|
1024 | =item db_family $family => $cb->(\%familyhash) |
|
|
1025 | |
|
|
1026 | Queries the named database C<$family> and call the callback with the |
|
|
1027 | family represented as a hash. You can keep and freely modify the hash. |
|
|
1028 | |
|
|
1029 | =item db_keys $family => $cb->(\@keys) |
|
|
1030 | |
|
|
1031 | Same as C<db_family>, except it only queries the family I<subkeys> and passes |
|
|
1032 | them as array reference to the callback. |
|
|
1033 | |
|
|
1034 | =item db_values $family => $cb->(\@values) |
|
|
1035 | |
|
|
1036 | Same as C<db_family>, except it only queries the family I<values> and passes them |
|
|
1037 | as array reference to the callback. |
|
|
1038 | |
|
|
1039 | =item $guard = db_mon $family => $cb->($familyhash, \@added, \@changed, \@deleted) |
|
|
1040 | |
|
|
1041 | Creates a monitor on the given database family. Each time a key is set |
|
|
1042 | or or is deleted the callback is called with a hash containing the |
|
|
1043 | database family and three lists of added, changed and deleted subkeys, |
|
|
1044 | respectively. If no keys have changed then the array reference might be |
|
|
1045 | C<undef> or even missing. |
|
|
1046 | |
|
|
1047 | If not called in void context, a guard object is returned that, when |
|
|
1048 | destroyed, stops the monitor. |
|
|
1049 | |
|
|
1050 | The family hash reference and the key arrays belong to AnyEvent::MP and |
|
|
1051 | B<must not be modified or stored> by the callback. When in doubt, make a |
|
|
1052 | copy. |
|
|
1053 | |
|
|
1054 | As soon as possible after the monitoring starts, the callback will be |
|
|
1055 | called with the intiial contents of the family, even if it is empty, |
|
|
1056 | i.e. there will always be a timely call to the callback with the current |
|
|
1057 | contents. |
|
|
1058 | |
|
|
1059 | It is possible that the callback is called with a change event even though |
|
|
1060 | the subkey is already present and the value has not changed. |
|
|
1061 | |
|
|
1062 | The monitoring stops when the guard object is destroyed. |
|
|
1063 | |
|
|
1064 | Example: on every change to the family "mygroup", print out all keys. |
|
|
1065 | |
|
|
1066 | my $guard = db_mon mygroup => sub { |
|
|
1067 | my ($family, $a, $c, $d) = @_; |
|
|
1068 | print "mygroup members: ", (join " ", keys %$family), "\n"; |
|
|
1069 | }; |
|
|
1070 | |
|
|
1071 | Exmaple: wait until the family "My::Module::workers" is non-empty. |
|
|
1072 | |
|
|
1073 | my $guard; $guard = db_mon My::Module::workers => sub { |
|
|
1074 | my ($family, $a, $c, $d) = @_; |
|
|
1075 | return unless %$family; |
|
|
1076 | undef $guard; |
|
|
1077 | print "My::Module::workers now nonempty\n"; |
|
|
1078 | }; |
|
|
1079 | |
|
|
1080 | Example: print all changes to the family "AnyRvent::Fantasy::Module". |
|
|
1081 | |
|
|
1082 | my $guard = db_mon AnyRvent::Fantasy::Module => sub { |
|
|
1083 | my ($family, $a, $c, $d) = @_; |
|
|
1084 | |
|
|
1085 | print "+$_=$family->{$_}\n" for @$a; |
|
|
1086 | print "*$_=$family->{$_}\n" for @$c; |
|
|
1087 | print "-$_=$family->{$_}\n" for @$d; |
|
|
1088 | }; |
|
|
1089 | |
|
|
1090 | =cut |
|
|
1091 | |
704 | =back |
1092 | =back |
705 | |
1093 | |
706 | =head1 AnyEvent::MP vs. Distributed Erlang |
1094 | =head1 AnyEvent::MP vs. Distributed Erlang |
707 | |
1095 | |
708 | AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node |
1096 | AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node |
709 | == aemp node, Erlang process == aemp port), so many of the documents and |
1097 | == aemp node, Erlang process == aemp port), so many of the documents and |
710 | programming techniques employed by Erlang apply to AnyEvent::MP. Here is a |
1098 | programming techniques employed by Erlang apply to AnyEvent::MP. Here is a |
711 | sample: |
1099 | sample: |
712 | |
1100 | |
713 | http://www.Erlang.se/doc/programming_rules.shtml |
1101 | http://www.erlang.se/doc/programming_rules.shtml |
714 | http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4 |
1102 | http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4 |
715 | http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6 |
1103 | http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6 |
716 | http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5 |
1104 | http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5 |
717 | |
1105 | |
718 | Despite the similarities, there are also some important differences: |
1106 | Despite the similarities, there are also some important differences: |
719 | |
1107 | |
720 | =over 4 |
1108 | =over 4 |
721 | |
1109 | |
722 | =item * Node IDs are arbitrary strings in AEMP. |
1110 | =item * Node IDs are arbitrary strings in AEMP. |
723 | |
1111 | |
724 | Erlang relies on special naming and DNS to work everywhere in the same |
1112 | Erlang relies on special naming and DNS to work everywhere in the same |
725 | way. AEMP relies on each node somehow knowing its own address(es) (e.g. by |
1113 | way. AEMP relies on each node somehow knowing its own address(es) (e.g. by |
726 | configuration or DNS), but will otherwise discover other odes itself. |
1114 | configuration or DNS), and possibly the addresses of some seed nodes, but |
|
|
1115 | will otherwise discover other nodes (and their IDs) itself. |
727 | |
1116 | |
728 | =item * Erlang has a "remote ports are like local ports" philosophy, AEMP |
1117 | =item * Erlang has a "remote ports are like local ports" philosophy, AEMP |
729 | uses "local ports are like remote ports". |
1118 | uses "local ports are like remote ports". |
730 | |
1119 | |
731 | The failure modes for local ports are quite different (runtime errors |
1120 | The failure modes for local ports are quite different (runtime errors |
… | |
… | |
740 | ports being the special case/exception, where transport errors cannot |
1129 | ports being the special case/exception, where transport errors cannot |
741 | occur. |
1130 | occur. |
742 | |
1131 | |
743 | =item * Erlang uses processes and a mailbox, AEMP does not queue. |
1132 | =item * Erlang uses processes and a mailbox, AEMP does not queue. |
744 | |
1133 | |
745 | Erlang uses processes that selectively receive messages, and therefore |
1134 | Erlang uses processes that selectively receive messages out of order, and |
746 | needs a queue. AEMP is event based, queuing messages would serve no |
1135 | therefore needs a queue. AEMP is event based, queuing messages would serve |
747 | useful purpose. For the same reason the pattern-matching abilities of |
1136 | no useful purpose. For the same reason the pattern-matching abilities |
748 | AnyEvent::MP are more limited, as there is little need to be able to |
1137 | of AnyEvent::MP are more limited, as there is little need to be able to |
749 | filter messages without dequeuing them. |
1138 | filter messages without dequeuing them. |
750 | |
1139 | |
751 | (But see L<Coro::MP> for a more Erlang-like process model on top of AEMP). |
1140 | This is not a philosophical difference, but simply stems from AnyEvent::MP |
|
|
1141 | being event-based, while Erlang is process-based. |
|
|
1142 | |
|
|
1143 | You cna have a look at L<Coro::MP> for a more Erlang-like process model on |
|
|
1144 | top of AEMP and Coro threads. |
752 | |
1145 | |
753 | =item * Erlang sends are synchronous, AEMP sends are asynchronous. |
1146 | =item * Erlang sends are synchronous, AEMP sends are asynchronous. |
754 | |
1147 | |
755 | Sending messages in Erlang is synchronous and blocks the process (and |
1148 | Sending messages in Erlang is synchronous and blocks the process until |
|
|
1149 | a conenction has been established and the message sent (and so does not |
756 | so does not need a queue that can overflow). AEMP sends are immediate, |
1150 | need a queue that can overflow). AEMP sends return immediately, connection |
757 | connection establishment is handled in the background. |
1151 | establishment is handled in the background. |
758 | |
1152 | |
759 | =item * Erlang suffers from silent message loss, AEMP does not. |
1153 | =item * Erlang suffers from silent message loss, AEMP does not. |
760 | |
1154 | |
761 | Erlang makes few guarantees on messages delivery - messages can get lost |
1155 | Erlang implements few guarantees on messages delivery - messages can get |
762 | without any of the processes realising it (i.e. you send messages a, b, |
1156 | lost without any of the processes realising it (i.e. you send messages a, |
763 | and c, and the other side only receives messages a and c). |
1157 | b, and c, and the other side only receives messages a and c). |
764 | |
1158 | |
765 | AEMP guarantees correct ordering, and the guarantee that after one message |
1159 | AEMP guarantees (modulo hardware errors) correct ordering, and the |
766 | is lost, all following ones sent to the same port are lost as well, until |
1160 | guarantee that after one message is lost, all following ones sent to the |
767 | monitoring raises an error, so there are no silent "holes" in the message |
1161 | same port are lost as well, until monitoring raises an error, so there are |
768 | sequence. |
1162 | no silent "holes" in the message sequence. |
|
|
1163 | |
|
|
1164 | If you want your software to be very reliable, you have to cope with |
|
|
1165 | corrupted and even out-of-order messages in both Erlang and AEMP. AEMP |
|
|
1166 | simply tries to work better in common error cases, such as when a network |
|
|
1167 | link goes down. |
769 | |
1168 | |
770 | =item * Erlang can send messages to the wrong port, AEMP does not. |
1169 | =item * Erlang can send messages to the wrong port, AEMP does not. |
771 | |
1170 | |
772 | In Erlang it is quite likely that a node that restarts reuses a process ID |
1171 | In Erlang it is quite likely that a node that restarts reuses an Erlang |
773 | known to other nodes for a completely different process, causing messages |
1172 | process ID known to other nodes for a completely different process, |
774 | destined for that process to end up in an unrelated process. |
1173 | causing messages destined for that process to end up in an unrelated |
|
|
1174 | process. |
775 | |
1175 | |
776 | AEMP never reuses port IDs, so old messages or old port IDs floating |
1176 | AEMP does not reuse port IDs, so old messages or old port IDs floating |
777 | around in the network will not be sent to an unrelated port. |
1177 | around in the network will not be sent to an unrelated port. |
778 | |
1178 | |
779 | =item * Erlang uses unprotected connections, AEMP uses secure |
1179 | =item * Erlang uses unprotected connections, AEMP uses secure |
780 | authentication and can use TLS. |
1180 | authentication and can use TLS. |
781 | |
1181 | |
… | |
… | |
784 | |
1184 | |
785 | =item * The AEMP protocol is optimised for both text-based and binary |
1185 | =item * The AEMP protocol is optimised for both text-based and binary |
786 | communications. |
1186 | communications. |
787 | |
1187 | |
788 | The AEMP protocol, unlike the Erlang protocol, supports both programming |
1188 | The AEMP protocol, unlike the Erlang protocol, supports both programming |
789 | language independent text-only protocols (good for debugging) and binary, |
1189 | language independent text-only protocols (good for debugging), and binary, |
790 | language-specific serialisers (e.g. Storable). By default, unless TLS is |
1190 | language-specific serialisers (e.g. Storable). By default, unless TLS is |
791 | used, the protocol is actually completely text-based. |
1191 | used, the protocol is actually completely text-based. |
792 | |
1192 | |
793 | It has also been carefully designed to be implementable in other languages |
1193 | It has also been carefully designed to be implementable in other languages |
794 | with a minimum of work while gracefully degrading functionality to make the |
1194 | with a minimum of work while gracefully degrading functionality to make the |
795 | protocol simple. |
1195 | protocol simple. |
796 | |
1196 | |
797 | =item * AEMP has more flexible monitoring options than Erlang. |
1197 | =item * AEMP has more flexible monitoring options than Erlang. |
798 | |
1198 | |
799 | In Erlang, you can chose to receive I<all> exit signals as messages |
1199 | In Erlang, you can chose to receive I<all> exit signals as messages or |
800 | or I<none>, there is no in-between, so monitoring single processes is |
1200 | I<none>, there is no in-between, so monitoring single Erlang processes is |
801 | difficult to implement. Monitoring in AEMP is more flexible than in |
1201 | difficult to implement. |
802 | Erlang, as one can choose between automatic kill, exit message or callback |
1202 | |
803 | on a per-process basis. |
1203 | Monitoring in AEMP is more flexible than in Erlang, as one can choose |
|
|
1204 | between automatic kill, exit message or callback on a per-port basis. |
804 | |
1205 | |
805 | =item * Erlang tries to hide remote/local connections, AEMP does not. |
1206 | =item * Erlang tries to hide remote/local connections, AEMP does not. |
806 | |
1207 | |
807 | Monitoring in Erlang is not an indicator of process death/crashes, in the |
1208 | Monitoring in Erlang is not an indicator of process death/crashes, in the |
808 | same way as linking is (except linking is unreliable in Erlang). |
1209 | same way as linking is (except linking is unreliable in Erlang). |
… | |
… | |
830 | overhead, as well as having to keep a proxy object everywhere. |
1231 | overhead, as well as having to keep a proxy object everywhere. |
831 | |
1232 | |
832 | Strings can easily be printed, easily serialised etc. and need no special |
1233 | Strings can easily be printed, easily serialised etc. and need no special |
833 | procedures to be "valid". |
1234 | procedures to be "valid". |
834 | |
1235 | |
835 | And as a result, a miniport consists of a single closure stored in a |
1236 | And as a result, a port with just a default receiver consists of a single |
836 | global hash - it can't become much cheaper. |
1237 | code reference stored in a global hash - it can't become much cheaper. |
837 | |
1238 | |
838 | =item Why favour JSON, why not a real serialising format such as Storable? |
1239 | =item Why favour JSON, why not a real serialising format such as Storable? |
839 | |
1240 | |
840 | In fact, any AnyEvent::MP node will happily accept Storable as framing |
1241 | In fact, any AnyEvent::MP node will happily accept Storable as framing |
841 | format, but currently there is no way to make a node use Storable by |
1242 | format, but currently there is no way to make a node use Storable by |
… | |
… | |
851 | Keeping your messages simple, concentrating on data structures rather than |
1252 | Keeping your messages simple, concentrating on data structures rather than |
852 | objects, will keep your messages clean, tidy and efficient. |
1253 | objects, will keep your messages clean, tidy and efficient. |
853 | |
1254 | |
854 | =back |
1255 | =back |
855 | |
1256 | |
|
|
1257 | =head1 PORTING FROM AnyEvent::MP VERSION 1.X |
|
|
1258 | |
|
|
1259 | AEMP version 2 has three major incompatible changes compared to version 1: |
|
|
1260 | |
|
|
1261 | =over 4 |
|
|
1262 | |
|
|
1263 | =item AnyEvent::MP::Global no longer has group management functions. |
|
|
1264 | |
|
|
1265 | AnyEvent::MP now comes with a distributed database that is more |
|
|
1266 | powerful. It's database families map closely to ports, but the API has |
|
|
1267 | minor differences: |
|
|
1268 | |
|
|
1269 | grp_reg $group, $port # old |
|
|
1270 | db_reg $group, $port # new |
|
|
1271 | |
|
|
1272 | $list = grp_get $group # old |
|
|
1273 | db_keys $group, sub { my $list = shift } # new |
|
|
1274 | |
|
|
1275 | grp_mon $group, $cb->(\@ports, $add, $del) # old |
|
|
1276 | db_mon $group, $cb->(\%ports, $add, $change, $del) # new |
|
|
1277 | |
|
|
1278 | C<grp_reg> is a no-brainer (just replace by C<db_reg>), but C<grp_get> |
|
|
1279 | is no longer instant, because the local node might not have a copy of |
|
|
1280 | the group. This can be partially remedied by using C<db_mon> to keep an |
|
|
1281 | updated copy of the group: |
|
|
1282 | |
|
|
1283 | my $local_group_copy; |
|
|
1284 | db_mon $group => sub { $local_group_copy = shift }; |
|
|
1285 | |
|
|
1286 | # no keys %$local_group_copy always returns the most up-to-date |
|
|
1287 | # list of ports in the group. |
|
|
1288 | |
|
|
1289 | C<grp_mon> can almost be replaced by C<db_mon>: |
|
|
1290 | |
|
|
1291 | db_mon $group => sub { |
|
|
1292 | my ($ports, $add, $chg, $lde) = @_; |
|
|
1293 | $ports = [keys %$ports]; |
|
|
1294 | |
|
|
1295 | # now $ports, $add and $del are the same as |
|
|
1296 | # were originally passed by grp_mon. |
|
|
1297 | ... |
|
|
1298 | }; |
|
|
1299 | |
|
|
1300 | =item Nodes not longer connect to all other nodes. |
|
|
1301 | |
|
|
1302 | In AEMP 1.x, every node automatically loads the L<AnyEvent::MP::Global> |
|
|
1303 | module, which in turn would create connections to all other nodes in the |
|
|
1304 | network (helped by the seed nodes). |
|
|
1305 | |
|
|
1306 | In version 2.x, global nodes still connect to all other global nodes, but |
|
|
1307 | other nodes don't - now every node either is a global node itself, or |
|
|
1308 | attaches itself to another global node. |
|
|
1309 | |
|
|
1310 | If a node isn't a global node itself, then it attaches itself to one |
|
|
1311 | of its seed nodes. If that seed node isn't a global node yet, it will |
|
|
1312 | automatically be upgraded to a global node. |
|
|
1313 | |
|
|
1314 | So in many cases, nothing needs to be changed - one just has to make sure |
|
|
1315 | that all seed nodes are meshed together with the other seed nodes (as with |
|
|
1316 | AEMP 1.x), and other nodes specify them as seed nodes. |
|
|
1317 | |
|
|
1318 | Not opening a connection to every other node is usually an advantage, |
|
|
1319 | except when you need the lower latency of an already established |
|
|
1320 | connection. To ensure a node establishes a connection to another node, |
|
|
1321 | you can monitor the node port (C<mon $node, ...>), which will attempt to |
|
|
1322 | create the connection (And notify you when the connection fails). |
|
|
1323 | |
|
|
1324 | =item Listener-less nodes are gone. |
|
|
1325 | |
|
|
1326 | And are not coming back, at least not in their old form. |
|
|
1327 | |
|
|
1328 | There are vague plans to implement some form of routing domains, which |
|
|
1329 | might or might not bring back listener-less nodes, but don't count on it. |
|
|
1330 | |
|
|
1331 | The fact that most connections are now optional somewhat mitigates this, |
|
|
1332 | as a node can be effectively unreachable from the outside without any |
|
|
1333 | problems, as long as it isn't a global node and only reaches out to other |
|
|
1334 | nodes (as opposed to being contacted from other nodes). |
|
|
1335 | |
|
|
1336 | =back |
|
|
1337 | |
856 | =head1 SEE ALSO |
1338 | =head1 SEE ALSO |
857 | |
1339 | |
858 | L<AnyEvent::MP::Intro> - a gentle introduction. |
1340 | L<AnyEvent::MP::Intro> - a gentle introduction. |
859 | |
1341 | |
860 | L<AnyEvent::MP::Kernel> - more, lower-level, stuff. |
1342 | L<AnyEvent::MP::Kernel> - more, lower-level, stuff. |
861 | |
1343 | |
862 | L<AnyEvent::MP::Global> - network maintainance and port groups, to find |
1344 | L<AnyEvent::MP::Global> - network maintenance and port groups, to find |
863 | your applications. |
1345 | your applications. |
|
|
1346 | |
|
|
1347 | L<AnyEvent::MP::DataConn> - establish data connections between nodes. |
864 | |
1348 | |
865 | L<AnyEvent::MP::LogCatcher> - simple service to display log messages from |
1349 | L<AnyEvent::MP::LogCatcher> - simple service to display log messages from |
866 | all nodes. |
1350 | all nodes. |
867 | |
1351 | |
868 | L<AnyEvent>. |
1352 | L<AnyEvent>. |