… | |
… | |
4 | |
4 | |
5 | =head1 SYNOPSIS |
5 | =head1 SYNOPSIS |
6 | |
6 | |
7 | use AnyEvent::MP; |
7 | use AnyEvent::MP; |
8 | |
8 | |
|
|
9 | $NODE # contains this node's node ID |
|
|
10 | NODE # returns this node's node ID |
|
|
11 | |
|
|
12 | $SELF # receiving/own port id in rcv callbacks |
|
|
13 | |
|
|
14 | # initialise the node so it can send/receive messages |
|
|
15 | configure; |
|
|
16 | |
|
|
17 | # ports are message destinations |
|
|
18 | |
|
|
19 | # sending messages |
|
|
20 | snd $port, type => data...; |
|
|
21 | snd $port, @msg; |
|
|
22 | snd @msg_with_first_element_being_a_port; |
|
|
23 | |
|
|
24 | # creating/using ports, the simple way |
|
|
25 | my $simple_port = port { my @msg = @_ }; |
|
|
26 | |
|
|
27 | # creating/using ports, tagged message matching |
|
|
28 | my $port = port; |
|
|
29 | rcv $port, ping => sub { snd $_[0], "pong" }; |
|
|
30 | rcv $port, pong => sub { warn "pong received\n" }; |
|
|
31 | |
|
|
32 | # create a port on another node |
|
|
33 | my $port = spawn $node, $initfunc, @initdata; |
|
|
34 | |
|
|
35 | # monitoring |
|
|
36 | mon $port, $cb->(@msg) # callback is invoked on death |
|
|
37 | mon $port, $otherport # kill otherport on abnormal death |
|
|
38 | mon $port, $otherport, @msg # send message on death |
|
|
39 | |
|
|
40 | =head1 CURRENT STATUS |
|
|
41 | |
|
|
42 | bin/aemp - stable. |
|
|
43 | AnyEvent::MP - stable API, should work. |
|
|
44 | AnyEvent::MP::Intro - explains most concepts. |
|
|
45 | AnyEvent::MP::Kernel - mostly stable. |
|
|
46 | AnyEvent::MP::Global - stable but incomplete, protocol not yet final. |
|
|
47 | |
|
|
48 | stay tuned. |
|
|
49 | |
9 | =head1 DESCRIPTION |
50 | =head1 DESCRIPTION |
10 | |
51 | |
|
|
52 | This module (-family) implements a simple message passing framework. |
|
|
53 | |
|
|
54 | Despite its simplicity, you can securely message other processes running |
|
|
55 | on the same or other hosts, and you can supervise entities remotely. |
|
|
56 | |
|
|
57 | For an introduction to this module family, see the L<AnyEvent::MP::Intro> |
|
|
58 | manual page and the examples under F<eg/>. |
|
|
59 | |
|
|
60 | =head1 CONCEPTS |
|
|
61 | |
|
|
62 | =over 4 |
|
|
63 | |
|
|
64 | =item port |
|
|
65 | |
|
|
66 | Not to be confused with a TCP port, a "port" is something you can send |
|
|
67 | messages to (with the C<snd> function). |
|
|
68 | |
|
|
69 | 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 |
|
|
71 | anything was listening for them or not. |
|
|
72 | |
|
|
73 | =item port ID - C<nodeid#portname> |
|
|
74 | |
|
|
75 | A port ID is the concatenation of a node ID, a hash-mark (C<#>) as |
|
|
76 | separator, and a port name (a printable string of unspecified format). |
|
|
77 | |
|
|
78 | =item node |
|
|
79 | |
|
|
80 | 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 |
|
|
82 | ports. |
|
|
83 | |
|
|
84 | Nodes are either public (have one or more listening ports) or private |
|
|
85 | (no listening ports). Private nodes cannot talk to other private nodes |
|
|
86 | currently. |
|
|
87 | |
|
|
88 | =item node ID - C<[A-Z_][a-zA-Z0-9_\-.:]*> |
|
|
89 | |
|
|
90 | 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 |
|
|
92 | hostname, a hostname and a port, or a random string. AnyEvent::MP itself |
|
|
93 | doesn't interpret node IDs in any way. |
|
|
94 | |
|
|
95 | =item binds - C<ip:port> |
|
|
96 | |
|
|
97 | 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 |
|
|
99 | endpoints - binds. Currently, only standard C<ip:port> specifications can |
|
|
100 | be used, which specify TCP ports to listen on. |
|
|
101 | |
|
|
102 | =item seed nodes |
|
|
103 | |
|
|
104 | When a node starts, it knows nothing about the network. To teach the node |
|
|
105 | about the network it first has to contact some other node within the |
|
|
106 | network. This node is called a seed. |
|
|
107 | |
|
|
108 | Apart from the fact that other nodes know them as seed nodes and they have |
|
|
109 | to have fixed listening addresses, seed nodes are perfectly normal nodes - |
|
|
110 | any node can function as a seed node for others. |
|
|
111 | |
|
|
112 | In addition to discovering the network, seed nodes are also used to |
|
|
113 | maintain the network and to connect nodes that otherwise would have |
|
|
114 | trouble connecting. They form the backbone of an AnyEvent::MP network. |
|
|
115 | |
|
|
116 | Seed nodes are expected to be long-running, and at least one seed node |
|
|
117 | should always be available. They should also be relatively responsive - a |
|
|
118 | seed node that blocks for long periods will slow down everybody else. |
|
|
119 | |
|
|
120 | =item seeds - C<host:port> |
|
|
121 | |
|
|
122 | Seeds are transport endpoint(s) (usually a hostname/IP address and a |
|
|
123 | TCP port) of nodes thta should be used as seed nodes. |
|
|
124 | |
|
|
125 | The nodes listening on those endpoints are expected to be long-running, |
|
|
126 | and at least one of those should always be available. When nodes run out |
|
|
127 | of connections (e.g. due to a network error), they try to re-establish |
|
|
128 | connections to some seednodes again to join the network. |
|
|
129 | |
|
|
130 | =back |
|
|
131 | |
|
|
132 | =head1 VARIABLES/FUNCTIONS |
|
|
133 | |
|
|
134 | =over 4 |
|
|
135 | |
11 | =cut |
136 | =cut |
12 | |
137 | |
13 | package AnyEvent::MP; |
138 | package AnyEvent::MP; |
14 | |
139 | |
|
|
140 | use AnyEvent::MP::Kernel; |
|
|
141 | |
15 | use common::sense; |
142 | use common::sense; |
16 | |
143 | |
|
|
144 | use Carp (); |
|
|
145 | |
17 | use AE (); |
146 | use AE (); |
18 | |
147 | |
19 | our $VERSION = '0.0'; |
148 | use base "Exporter"; |
20 | |
149 | |
21 | sub nonce($) { |
150 | our $VERSION = $AnyEvent::MP::Kernel::VERSION; |
22 | my $nonce; |
|
|
23 | |
151 | |
24 | if (open my $fh, "</dev/urandom") { |
152 | our @EXPORT = qw( |
25 | sysread $fh, $nonce, $_[0]; |
153 | NODE $NODE *SELF node_of after |
26 | } else { |
154 | configure |
|
|
155 | snd rcv mon mon_guard kil reg psub spawn |
|
|
156 | port |
|
|
157 | ); |
|
|
158 | |
|
|
159 | our $SELF; |
|
|
160 | |
|
|
161 | sub _self_die() { |
|
|
162 | my $msg = $@; |
|
|
163 | $msg =~ s/\n+$// unless ref $msg; |
|
|
164 | kil $SELF, die => $msg; |
|
|
165 | } |
|
|
166 | |
|
|
167 | =item $thisnode = NODE / $NODE |
|
|
168 | |
|
|
169 | The C<NODE> function returns, and the C<$NODE> variable contains, the node |
|
|
170 | ID of the node running in the current process. This value is initialised by |
|
|
171 | a call to C<configure>. |
|
|
172 | |
|
|
173 | =item $nodeid = node_of $port |
|
|
174 | |
|
|
175 | Extracts and returns the node ID from a port ID or a node ID. |
|
|
176 | |
|
|
177 | =item configure $profile, key => value... |
|
|
178 | |
|
|
179 | =item configure key => value... |
|
|
180 | |
|
|
181 | Before a node can talk to other nodes on the network (i.e. enter |
|
|
182 | "distributed mode") it has to configure itself - the minimum a node needs |
|
|
183 | to know is its own name, and optionally it should know the addresses of |
|
|
184 | some other nodes in the network to discover other nodes. |
|
|
185 | |
|
|
186 | This function configures a node - it must be called exactly once (or |
|
|
187 | never) before calling other AnyEvent::MP functions. |
|
|
188 | |
|
|
189 | =over 4 |
|
|
190 | |
|
|
191 | =item step 1, gathering configuration from profiles |
|
|
192 | |
|
|
193 | The function first looks up a profile in the aemp configuration (see the |
|
|
194 | L<aemp> commandline utility). The profile name can be specified via the |
|
|
195 | named C<profile> parameter or can simply be the first parameter). If it is |
|
|
196 | missing, then the nodename (F<uname -n>) will be used as profile name. |
|
|
197 | |
|
|
198 | The profile data is then gathered as follows: |
|
|
199 | |
|
|
200 | First, all remaining key => value pairs (all of which are conveniently |
|
|
201 | undocumented at the moment) will be interpreted as configuration |
|
|
202 | data. Then they will be overwritten by any values specified in the global |
|
|
203 | default configuration (see the F<aemp> utility), then the chain of |
|
|
204 | profiles chosen by the profile name (and any C<parent> attributes). |
|
|
205 | |
|
|
206 | That means that the values specified in the profile have highest priority |
|
|
207 | and the values specified directly via C<configure> have lowest priority, |
|
|
208 | and can only be used to specify defaults. |
|
|
209 | |
|
|
210 | If the profile specifies a node ID, then this will become the node ID of |
|
|
211 | this process. If not, then the profile name will be used as node ID. The |
|
|
212 | special node ID of C<anon/> will be replaced by a random node ID. |
|
|
213 | |
|
|
214 | =item step 2, bind listener sockets |
|
|
215 | |
|
|
216 | The next step is to look up the binds in the profile, followed by binding |
|
|
217 | aemp protocol listeners on all binds specified (it is possible and valid |
|
|
218 | to have no binds, meaning that the node cannot be contacted form the |
|
|
219 | outside. This means the node cannot talk to other nodes that also have no |
|
|
220 | binds, but it can still talk to all "normal" nodes). |
|
|
221 | |
|
|
222 | If the profile does not specify a binds list, then a default of C<*> is |
|
|
223 | used, meaning the node will bind on a dynamically-assigned port on every |
|
|
224 | local IP address it finds. |
|
|
225 | |
|
|
226 | =item step 3, connect to seed nodes |
|
|
227 | |
|
|
228 | As the last step, the seeds list from the profile is passed to the |
|
|
229 | L<AnyEvent::MP::Global> module, which will then use it to keep |
|
|
230 | connectivity with at least one node at any point in time. |
|
|
231 | |
|
|
232 | =back |
|
|
233 | |
|
|
234 | Example: become a distributed node using the locla node name as profile. |
|
|
235 | This should be the most common form of invocation for "daemon"-type nodes. |
|
|
236 | |
|
|
237 | configure |
|
|
238 | |
|
|
239 | Example: become an anonymous node. This form is often used for commandline |
|
|
240 | clients. |
|
|
241 | |
|
|
242 | configure nodeid => "anon/"; |
|
|
243 | |
|
|
244 | Example: configure a node using a profile called seed, which si suitable |
|
|
245 | for a seed node as it binds on all local addresses on a fixed port (4040, |
|
|
246 | customary for aemp). |
|
|
247 | |
|
|
248 | # use the aemp commandline utility |
|
|
249 | # aemp profile seed nodeid anon/ binds '*:4040' |
|
|
250 | |
|
|
251 | # then use it |
|
|
252 | configure profile => "seed"; |
|
|
253 | |
|
|
254 | # or simply use aemp from the shell again: |
|
|
255 | # aemp run profile seed |
|
|
256 | |
|
|
257 | # or provide a nicer-to-remember nodeid |
|
|
258 | # aemp run profile seed nodeid "$(hostname)" |
|
|
259 | |
|
|
260 | =item $SELF |
|
|
261 | |
|
|
262 | Contains the current port id while executing C<rcv> callbacks or C<psub> |
|
|
263 | blocks. |
|
|
264 | |
|
|
265 | =item *SELF, SELF, %SELF, @SELF... |
|
|
266 | |
|
|
267 | Due to some quirks in how perl exports variables, it is impossible to |
|
|
268 | just export C<$SELF>, all the symbols named C<SELF> are exported by this |
|
|
269 | module, but only C<$SELF> is currently used. |
|
|
270 | |
|
|
271 | =item snd $port, type => @data |
|
|
272 | |
|
|
273 | =item snd $port, @msg |
|
|
274 | |
|
|
275 | Send the given message to the given port, which can identify either a |
|
|
276 | local or a remote port, and must be a port ID. |
|
|
277 | |
|
|
278 | While the message can be almost anything, it is highly recommended to |
|
|
279 | use a string as first element (a port ID, or some word that indicates a |
|
|
280 | request type etc.) and to consist if only simple perl values (scalars, |
|
|
281 | arrays, hashes) - if you think you need to pass an object, think again. |
|
|
282 | |
|
|
283 | The message data logically becomes read-only after a call to this |
|
|
284 | function: modifying any argument (or values referenced by them) is |
|
|
285 | forbidden, as there can be considerable time between the call to C<snd> |
|
|
286 | and the time the message is actually being serialised - in fact, it might |
|
|
287 | never be copied as within the same process it is simply handed to the |
|
|
288 | receiving port. |
|
|
289 | |
|
|
290 | The type of data you can transfer depends on the transport protocol: when |
|
|
291 | JSON is used, then only strings, numbers and arrays and hashes consisting |
|
|
292 | of those are allowed (no objects). When Storable is used, then anything |
|
|
293 | that Storable can serialise and deserialise is allowed, and for the local |
|
|
294 | node, anything can be passed. Best rely only on the common denominator of |
|
|
295 | these. |
|
|
296 | |
|
|
297 | =item $local_port = port |
|
|
298 | |
|
|
299 | Create a new local port object and returns its port ID. Initially it has |
|
|
300 | no callbacks set and will throw an error when it receives messages. |
|
|
301 | |
|
|
302 | =item $local_port = port { my @msg = @_ } |
|
|
303 | |
|
|
304 | Creates a new local port, and returns its ID. Semantically the same as |
|
|
305 | creating a port and calling C<rcv $port, $callback> on it. |
|
|
306 | |
|
|
307 | The block will be called for every message received on the port, with the |
|
|
308 | global variable C<$SELF> set to the port ID. Runtime errors will cause the |
|
|
309 | port to be C<kil>ed. The message will be passed as-is, no extra argument |
|
|
310 | (i.e. no port ID) will be passed to the callback. |
|
|
311 | |
|
|
312 | If you want to stop/destroy the port, simply C<kil> it: |
|
|
313 | |
|
|
314 | my $port = port { |
|
|
315 | my @msg = @_; |
27 | # shit... |
316 | ... |
28 | our $nonce_init; |
317 | kil $SELF; |
29 | unless ($nonce_init++) { |
|
|
30 | srand time ^ $$ ^ unpack "%L*", qx"ps -edalf" . qx"ipconfig /all"; |
|
|
31 | } |
|
|
32 | |
|
|
33 | $nonce = join "", map +(chr rand 256), 1 .. $_[0] |
|
|
34 | } |
318 | }; |
35 | |
319 | |
36 | $nonce |
320 | =cut |
37 | } |
|
|
38 | |
321 | |
39 | our $DEFAULT_SECRET; |
322 | sub rcv($@); |
40 | |
323 | |
41 | sub default_secret { |
324 | sub _kilme { |
42 | unless (defined $DEFAULT_SECRET) { |
325 | die "received message on port without callback"; |
43 | if (open my $fh, "<$ENV{HOME}/.aemp-secret") { |
326 | } |
44 | sysread $fh, $DEFAULT_SECRET, -s $fh; |
327 | |
|
|
328 | sub port(;&) { |
|
|
329 | my $id = "$UNIQ." . $ID++; |
|
|
330 | my $port = "$NODE#$id"; |
|
|
331 | |
|
|
332 | rcv $port, shift || \&_kilme; |
|
|
333 | |
|
|
334 | $port |
|
|
335 | } |
|
|
336 | |
|
|
337 | =item rcv $local_port, $callback->(@msg) |
|
|
338 | |
|
|
339 | Replaces the default callback on the specified port. There is no way to |
|
|
340 | remove the default callback: use C<sub { }> to disable it, or better |
|
|
341 | C<kil> the port when it is no longer needed. |
|
|
342 | |
|
|
343 | The global C<$SELF> (exported by this module) contains C<$port> while |
|
|
344 | executing the callback. Runtime errors during callback execution will |
|
|
345 | result in the port being C<kil>ed. |
|
|
346 | |
|
|
347 | The default callback received all messages not matched by a more specific |
|
|
348 | C<tag> match. |
|
|
349 | |
|
|
350 | =item rcv $local_port, tag => $callback->(@msg_without_tag), ... |
|
|
351 | |
|
|
352 | Register (or replace) callbacks to be called on messages starting with the |
|
|
353 | given tag on the given port (and return the port), or unregister it (when |
|
|
354 | C<$callback> is C<$undef> or missing). There can only be one callback |
|
|
355 | registered for each tag. |
|
|
356 | |
|
|
357 | The original message will be passed to the callback, after the first |
|
|
358 | element (the tag) has been removed. The callback will use the same |
|
|
359 | environment as the default callback (see above). |
|
|
360 | |
|
|
361 | Example: create a port and bind receivers on it in one go. |
|
|
362 | |
|
|
363 | my $port = rcv port, |
|
|
364 | msg1 => sub { ... }, |
|
|
365 | msg2 => sub { ... }, |
|
|
366 | ; |
|
|
367 | |
|
|
368 | Example: create a port, bind receivers and send it in a message elsewhere |
|
|
369 | in one go: |
|
|
370 | |
|
|
371 | snd $otherport, reply => |
|
|
372 | rcv port, |
|
|
373 | msg1 => sub { ... }, |
|
|
374 | ... |
|
|
375 | ; |
|
|
376 | |
|
|
377 | Example: temporarily register a rcv callback for a tag matching some port |
|
|
378 | (e.g. for a rpc reply) and unregister it after a message was received. |
|
|
379 | |
|
|
380 | rcv $port, $otherport => sub { |
|
|
381 | my @reply = @_; |
|
|
382 | |
|
|
383 | rcv $SELF, $otherport; |
|
|
384 | }; |
|
|
385 | |
|
|
386 | =cut |
|
|
387 | |
|
|
388 | sub rcv($@) { |
|
|
389 | my $port = shift; |
|
|
390 | my ($nodeid, $portid) = split /#/, $port, 2; |
|
|
391 | |
|
|
392 | $NODE{$nodeid} == $NODE{""} |
|
|
393 | or Carp::croak "$port: rcv can only be called on local ports, caught"; |
|
|
394 | |
|
|
395 | while (@_) { |
|
|
396 | if (ref $_[0]) { |
|
|
397 | if (my $self = $PORT_DATA{$portid}) { |
|
|
398 | "AnyEvent::MP::Port" eq ref $self |
|
|
399 | or Carp::croak "$port: rcv can only be called on message matching ports, caught"; |
|
|
400 | |
|
|
401 | $self->[2] = shift; |
45 | } else { |
402 | } else { |
46 | $DEFAULT_SECRET = nonce 32; |
403 | my $cb = shift; |
|
|
404 | $PORT{$portid} = sub { |
|
|
405 | local $SELF = $port; |
|
|
406 | eval { &$cb }; _self_die if $@; |
|
|
407 | }; |
|
|
408 | } |
|
|
409 | } elsif (defined $_[0]) { |
|
|
410 | my $self = $PORT_DATA{$portid} ||= do { |
|
|
411 | my $self = bless [$PORT{$port} || sub { }, { }, $port], "AnyEvent::MP::Port"; |
|
|
412 | |
|
|
413 | $PORT{$portid} = sub { |
|
|
414 | local $SELF = $port; |
|
|
415 | |
|
|
416 | if (my $cb = $self->[1]{$_[0]}) { |
|
|
417 | shift; |
|
|
418 | eval { &$cb }; _self_die if $@; |
|
|
419 | } else { |
|
|
420 | &{ $self->[0] }; |
|
|
421 | } |
|
|
422 | }; |
|
|
423 | |
|
|
424 | $self |
|
|
425 | }; |
|
|
426 | |
|
|
427 | "AnyEvent::MP::Port" eq ref $self |
|
|
428 | or Carp::croak "$port: rcv can only be called on message matching ports, caught"; |
|
|
429 | |
|
|
430 | my ($tag, $cb) = splice @_, 0, 2; |
|
|
431 | |
|
|
432 | if (defined $cb) { |
|
|
433 | $self->[1]{$tag} = $cb; |
|
|
434 | } else { |
|
|
435 | delete $self->[1]{$tag}; |
|
|
436 | } |
47 | } |
437 | } |
48 | } |
438 | } |
49 | |
439 | |
50 | $DEFAULT_SECRET |
440 | $port |
51 | } |
441 | } |
|
|
442 | |
|
|
443 | =item $closure = psub { BLOCK } |
|
|
444 | |
|
|
445 | Remembers C<$SELF> and creates a closure out of the BLOCK. When the |
|
|
446 | closure is executed, sets up the environment in the same way as in C<rcv> |
|
|
447 | callbacks, i.e. runtime errors will cause the port to get C<kil>ed. |
|
|
448 | |
|
|
449 | This is useful when you register callbacks from C<rcv> callbacks: |
|
|
450 | |
|
|
451 | rcv delayed_reply => sub { |
|
|
452 | my ($delay, @reply) = @_; |
|
|
453 | my $timer = AE::timer $delay, 0, psub { |
|
|
454 | snd @reply, $SELF; |
|
|
455 | }; |
|
|
456 | }; |
|
|
457 | |
|
|
458 | =cut |
|
|
459 | |
|
|
460 | sub psub(&) { |
|
|
461 | my $cb = shift; |
|
|
462 | |
|
|
463 | my $port = $SELF |
|
|
464 | or Carp::croak "psub can only be called from within rcv or psub callbacks, not"; |
|
|
465 | |
|
|
466 | sub { |
|
|
467 | local $SELF = $port; |
|
|
468 | |
|
|
469 | if (wantarray) { |
|
|
470 | my @res = eval { &$cb }; |
|
|
471 | _self_die if $@; |
|
|
472 | @res |
|
|
473 | } else { |
|
|
474 | my $res = eval { &$cb }; |
|
|
475 | _self_die if $@; |
|
|
476 | $res |
|
|
477 | } |
|
|
478 | } |
|
|
479 | } |
|
|
480 | |
|
|
481 | =item $guard = mon $port, $cb->(@reason) # call $cb when $port dies |
|
|
482 | |
|
|
483 | =item $guard = mon $port, $rcvport # kill $rcvport when $port dies |
|
|
484 | |
|
|
485 | =item $guard = mon $port # kill $SELF when $port dies |
|
|
486 | |
|
|
487 | =item $guard = mon $port, $rcvport, @msg # send a message when $port dies |
|
|
488 | |
|
|
489 | Monitor the given port and do something when the port is killed or |
|
|
490 | messages to it were lost, and optionally return a guard that can be used |
|
|
491 | to stop monitoring again. |
|
|
492 | |
|
|
493 | In the first form (callback), the callback is simply called with any |
|
|
494 | number of C<@reason> elements (no @reason means that the port was deleted |
|
|
495 | "normally"). Note also that I<< the callback B<must> never die >>, so use |
|
|
496 | C<eval> if unsure. |
|
|
497 | |
|
|
498 | In the second form (another port given), the other port (C<$rcvport>) |
|
|
499 | will be C<kil>'ed with C<@reason>, if a @reason was specified, i.e. on |
|
|
500 | "normal" kils nothing happens, while under all other conditions, the other |
|
|
501 | port is killed with the same reason. |
|
|
502 | |
|
|
503 | The third form (kill self) is the same as the second form, except that |
|
|
504 | C<$rvport> defaults to C<$SELF>. |
|
|
505 | |
|
|
506 | In the last form (message), a message of the form C<@msg, @reason> will be |
|
|
507 | C<snd>. |
|
|
508 | |
|
|
509 | Monitoring-actions are one-shot: once messages are lost (and a monitoring |
|
|
510 | alert was raised), they are removed and will not trigger again. |
|
|
511 | |
|
|
512 | As a rule of thumb, monitoring requests should always monitor a port from |
|
|
513 | a local port (or callback). The reason is that kill messages might get |
|
|
514 | lost, just like any other message. Another less obvious reason is that |
|
|
515 | even monitoring requests can get lost (for example, when the connection |
|
|
516 | to the other node goes down permanently). When monitoring a port locally |
|
|
517 | these problems do not exist. |
|
|
518 | |
|
|
519 | C<mon> effectively guarantees that, in the absence of hardware failures, |
|
|
520 | after starting the monitor, either all messages sent to the port will |
|
|
521 | arrive, or the monitoring action will be invoked after possible message |
|
|
522 | loss has been detected. No messages will be lost "in between" (after |
|
|
523 | the first lost message no further messages will be received by the |
|
|
524 | port). After the monitoring action was invoked, further messages might get |
|
|
525 | delivered again. |
|
|
526 | |
|
|
527 | Inter-host-connection timeouts and monitoring depend on the transport |
|
|
528 | used. The only transport currently implemented is TCP, and AnyEvent::MP |
|
|
529 | relies on TCP to detect node-downs (this can take 10-15 minutes on a |
|
|
530 | non-idle connection, and usually around two hours for idle conenctions). |
|
|
531 | |
|
|
532 | This means that monitoring is good for program errors and cleaning up |
|
|
533 | stuff eventually, but they are no replacement for a timeout when you need |
|
|
534 | to ensure some maximum latency. |
|
|
535 | |
|
|
536 | Example: call a given callback when C<$port> is killed. |
|
|
537 | |
|
|
538 | mon $port, sub { warn "port died because of <@_>\n" }; |
|
|
539 | |
|
|
540 | Example: kill ourselves when C<$port> is killed abnormally. |
|
|
541 | |
|
|
542 | mon $port; |
|
|
543 | |
|
|
544 | Example: send us a restart message when another C<$port> is killed. |
|
|
545 | |
|
|
546 | mon $port, $self => "restart"; |
|
|
547 | |
|
|
548 | =cut |
|
|
549 | |
|
|
550 | sub mon { |
|
|
551 | my ($nodeid, $port) = split /#/, shift, 2; |
|
|
552 | |
|
|
553 | my $node = $NODE{$nodeid} || add_node $nodeid; |
|
|
554 | |
|
|
555 | my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,'; |
|
|
556 | |
|
|
557 | unless (ref $cb) { |
|
|
558 | if (@_) { |
|
|
559 | # send a kill info message |
|
|
560 | my (@msg) = ($cb, @_); |
|
|
561 | $cb = sub { snd @msg, @_ }; |
|
|
562 | } else { |
|
|
563 | # simply kill other port |
|
|
564 | my $port = $cb; |
|
|
565 | $cb = sub { kil $port, @_ if @_ }; |
|
|
566 | } |
|
|
567 | } |
|
|
568 | |
|
|
569 | $node->monitor ($port, $cb); |
|
|
570 | |
|
|
571 | defined wantarray |
|
|
572 | and AnyEvent::Util::guard { $node->unmonitor ($port, $cb) } |
|
|
573 | } |
|
|
574 | |
|
|
575 | =item $guard = mon_guard $port, $ref, $ref... |
|
|
576 | |
|
|
577 | Monitors the given C<$port> and keeps the passed references. When the port |
|
|
578 | is killed, the references will be freed. |
|
|
579 | |
|
|
580 | Optionally returns a guard that will stop the monitoring. |
|
|
581 | |
|
|
582 | This function is useful when you create e.g. timers or other watchers and |
|
|
583 | want to free them when the port gets killed (note the use of C<psub>): |
|
|
584 | |
|
|
585 | $port->rcv (start => sub { |
|
|
586 | my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub { |
|
|
587 | undef $timer if 0.9 < rand; |
|
|
588 | }); |
|
|
589 | }); |
|
|
590 | |
|
|
591 | =cut |
|
|
592 | |
|
|
593 | sub mon_guard { |
|
|
594 | my ($port, @refs) = @_; |
|
|
595 | |
|
|
596 | #TODO: mon-less form? |
|
|
597 | |
|
|
598 | mon $port, sub { 0 && @refs } |
|
|
599 | } |
|
|
600 | |
|
|
601 | =item kil $port[, @reason] |
|
|
602 | |
|
|
603 | Kill the specified port with the given C<@reason>. |
|
|
604 | |
|
|
605 | If no C<@reason> is specified, then the port is killed "normally" (ports |
|
|
606 | monitoring other ports will not necessarily die because a port dies |
|
|
607 | "normally"). |
|
|
608 | |
|
|
609 | Otherwise, linked ports get killed with the same reason (second form of |
|
|
610 | C<mon>, see above). |
|
|
611 | |
|
|
612 | Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks |
|
|
613 | will be reported as reason C<< die => $@ >>. |
|
|
614 | |
|
|
615 | Transport/communication errors are reported as C<< transport_error => |
|
|
616 | $message >>. |
|
|
617 | |
|
|
618 | =cut |
|
|
619 | |
|
|
620 | =item $port = spawn $node, $initfunc[, @initdata] |
|
|
621 | |
|
|
622 | Creates a port on the node C<$node> (which can also be a port ID, in which |
|
|
623 | case it's the node where that port resides). |
|
|
624 | |
|
|
625 | The port ID of the newly created port is returned immediately, and it is |
|
|
626 | possible to immediately start sending messages or to monitor the port. |
|
|
627 | |
|
|
628 | After the port has been created, the init function is called on the remote |
|
|
629 | node, in the same context as a C<rcv> callback. This function must be a |
|
|
630 | fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To |
|
|
631 | specify a function in the main program, use C<::name>. |
|
|
632 | |
|
|
633 | If the function doesn't exist, then the node tries to C<require> |
|
|
634 | the package, then the package above the package and so on (e.g. |
|
|
635 | C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function |
|
|
636 | exists or it runs out of package names. |
|
|
637 | |
|
|
638 | The init function is then called with the newly-created port as context |
|
|
639 | object (C<$SELF>) and the C<@initdata> values as arguments. It I<must> |
|
|
640 | call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise |
|
|
641 | the port might not get created. |
|
|
642 | |
|
|
643 | A common idiom is to pass a local port, immediately monitor the spawned |
|
|
644 | port, and in the remote init function, immediately monitor the passed |
|
|
645 | local port. This two-way monitoring ensures that both ports get cleaned up |
|
|
646 | when there is a problem. |
|
|
647 | |
|
|
648 | C<spawn> guarantees that the C<$initfunc> has no visible effects on the |
|
|
649 | caller before C<spawn> returns (by delaying invocation when spawn is |
|
|
650 | called for the local node). |
|
|
651 | |
|
|
652 | Example: spawn a chat server port on C<$othernode>. |
|
|
653 | |
|
|
654 | # this node, executed from within a port context: |
|
|
655 | my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF; |
|
|
656 | mon $server; |
|
|
657 | |
|
|
658 | # init function on C<$othernode> |
|
|
659 | sub connect { |
|
|
660 | my ($srcport) = @_; |
|
|
661 | |
|
|
662 | mon $srcport; |
|
|
663 | |
|
|
664 | rcv $SELF, sub { |
|
|
665 | ... |
|
|
666 | }; |
|
|
667 | } |
|
|
668 | |
|
|
669 | =cut |
|
|
670 | |
|
|
671 | sub _spawn { |
|
|
672 | my $port = shift; |
|
|
673 | my $init = shift; |
|
|
674 | |
|
|
675 | # rcv will create the actual port |
|
|
676 | local $SELF = "$NODE#$port"; |
|
|
677 | eval { |
|
|
678 | &{ load_func $init } |
|
|
679 | }; |
|
|
680 | _self_die if $@; |
|
|
681 | } |
|
|
682 | |
|
|
683 | sub spawn(@) { |
|
|
684 | my ($nodeid, undef) = split /#/, shift, 2; |
|
|
685 | |
|
|
686 | my $id = "$RUNIQ." . $ID++; |
|
|
687 | |
|
|
688 | $_[0] =~ /::/ |
|
|
689 | or Carp::croak "spawn init function must be a fully-qualified name, caught"; |
|
|
690 | |
|
|
691 | snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_; |
|
|
692 | |
|
|
693 | "$nodeid#$id" |
|
|
694 | } |
|
|
695 | |
|
|
696 | =item after $timeout, @msg |
|
|
697 | |
|
|
698 | =item after $timeout, $callback |
|
|
699 | |
|
|
700 | Either sends the given message, or call the given callback, after the |
|
|
701 | specified number of seconds. |
|
|
702 | |
|
|
703 | This is simply a utility function that comes in handy at times - the |
|
|
704 | AnyEvent::MP author is not convinced of the wisdom of having it, though, |
|
|
705 | so it may go away in the future. |
|
|
706 | |
|
|
707 | =cut |
|
|
708 | |
|
|
709 | sub after($@) { |
|
|
710 | my ($timeout, @action) = @_; |
|
|
711 | |
|
|
712 | my $t; $t = AE::timer $timeout, 0, sub { |
|
|
713 | undef $t; |
|
|
714 | ref $action[0] |
|
|
715 | ? $action[0]() |
|
|
716 | : snd @action; |
|
|
717 | }; |
|
|
718 | } |
|
|
719 | |
|
|
720 | =back |
|
|
721 | |
|
|
722 | =head1 AnyEvent::MP vs. Distributed Erlang |
|
|
723 | |
|
|
724 | AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node |
|
|
725 | == aemp node, Erlang process == aemp port), so many of the documents and |
|
|
726 | programming techniques employed by Erlang apply to AnyEvent::MP. Here is a |
|
|
727 | sample: |
|
|
728 | |
|
|
729 | http://www.Erlang.se/doc/programming_rules.shtml |
|
|
730 | http://Erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4 |
|
|
731 | http://Erlang.org/download/Erlang-book-part1.pdf # chapters 5 and 6 |
|
|
732 | http://Erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5 |
|
|
733 | |
|
|
734 | Despite the similarities, there are also some important differences: |
|
|
735 | |
|
|
736 | =over 4 |
|
|
737 | |
|
|
738 | =item * Node IDs are arbitrary strings in AEMP. |
|
|
739 | |
|
|
740 | Erlang relies on special naming and DNS to work everywhere in the same |
|
|
741 | way. AEMP relies on each node somehow knowing its own address(es) (e.g. by |
|
|
742 | configuration or DNS), but will otherwise discover other odes itself. |
|
|
743 | |
|
|
744 | =item * Erlang has a "remote ports are like local ports" philosophy, AEMP |
|
|
745 | uses "local ports are like remote ports". |
|
|
746 | |
|
|
747 | The failure modes for local ports are quite different (runtime errors |
|
|
748 | only) then for remote ports - when a local port dies, you I<know> it dies, |
|
|
749 | when a connection to another node dies, you know nothing about the other |
|
|
750 | port. |
|
|
751 | |
|
|
752 | Erlang pretends remote ports are as reliable as local ports, even when |
|
|
753 | they are not. |
|
|
754 | |
|
|
755 | AEMP encourages a "treat remote ports differently" philosophy, with local |
|
|
756 | ports being the special case/exception, where transport errors cannot |
|
|
757 | occur. |
|
|
758 | |
|
|
759 | =item * Erlang uses processes and a mailbox, AEMP does not queue. |
|
|
760 | |
|
|
761 | Erlang uses processes that selectively receive messages, and therefore |
|
|
762 | needs a queue. AEMP is event based, queuing messages would serve no |
|
|
763 | useful purpose. For the same reason the pattern-matching abilities of |
|
|
764 | AnyEvent::MP are more limited, as there is little need to be able to |
|
|
765 | filter messages without dequeuing them. |
|
|
766 | |
|
|
767 | (But see L<Coro::MP> for a more Erlang-like process model on top of AEMP). |
|
|
768 | |
|
|
769 | =item * Erlang sends are synchronous, AEMP sends are asynchronous. |
|
|
770 | |
|
|
771 | Sending messages in Erlang is synchronous and blocks the process (and |
|
|
772 | so does not need a queue that can overflow). AEMP sends are immediate, |
|
|
773 | connection establishment is handled in the background. |
|
|
774 | |
|
|
775 | =item * Erlang suffers from silent message loss, AEMP does not. |
|
|
776 | |
|
|
777 | Erlang makes few guarantees on messages delivery - messages can get lost |
|
|
778 | without any of the processes realising it (i.e. you send messages a, b, |
|
|
779 | and c, and the other side only receives messages a and c). |
|
|
780 | |
|
|
781 | AEMP guarantees correct ordering, and the guarantee that after one message |
|
|
782 | is lost, all following ones sent to the same port are lost as well, until |
|
|
783 | monitoring raises an error, so there are no silent "holes" in the message |
|
|
784 | sequence. |
|
|
785 | |
|
|
786 | =item * Erlang can send messages to the wrong port, AEMP does not. |
|
|
787 | |
|
|
788 | In Erlang it is quite likely that a node that restarts reuses a process ID |
|
|
789 | known to other nodes for a completely different process, causing messages |
|
|
790 | destined for that process to end up in an unrelated process. |
|
|
791 | |
|
|
792 | AEMP never reuses port IDs, so old messages or old port IDs floating |
|
|
793 | around in the network will not be sent to an unrelated port. |
|
|
794 | |
|
|
795 | =item * Erlang uses unprotected connections, AEMP uses secure |
|
|
796 | authentication and can use TLS. |
|
|
797 | |
|
|
798 | AEMP can use a proven protocol - TLS - to protect connections and |
|
|
799 | securely authenticate nodes. |
|
|
800 | |
|
|
801 | =item * The AEMP protocol is optimised for both text-based and binary |
|
|
802 | communications. |
|
|
803 | |
|
|
804 | The AEMP protocol, unlike the Erlang protocol, supports both programming |
|
|
805 | language independent text-only protocols (good for debugging) and binary, |
|
|
806 | language-specific serialisers (e.g. Storable). By default, unless TLS is |
|
|
807 | used, the protocol is actually completely text-based. |
|
|
808 | |
|
|
809 | It has also been carefully designed to be implementable in other languages |
|
|
810 | with a minimum of work while gracefully degrading functionality to make the |
|
|
811 | protocol simple. |
|
|
812 | |
|
|
813 | =item * AEMP has more flexible monitoring options than Erlang. |
|
|
814 | |
|
|
815 | In Erlang, you can chose to receive I<all> exit signals as messages |
|
|
816 | or I<none>, there is no in-between, so monitoring single processes is |
|
|
817 | difficult to implement. Monitoring in AEMP is more flexible than in |
|
|
818 | Erlang, as one can choose between automatic kill, exit message or callback |
|
|
819 | on a per-process basis. |
|
|
820 | |
|
|
821 | =item * Erlang tries to hide remote/local connections, AEMP does not. |
|
|
822 | |
|
|
823 | Monitoring in Erlang is not an indicator of process death/crashes, in the |
|
|
824 | same way as linking is (except linking is unreliable in Erlang). |
|
|
825 | |
|
|
826 | In AEMP, you don't "look up" registered port names or send to named ports |
|
|
827 | that might or might not be persistent. Instead, you normally spawn a port |
|
|
828 | on the remote node. The init function monitors you, and you monitor the |
|
|
829 | remote port. Since both monitors are local to the node, they are much more |
|
|
830 | reliable (no need for C<spawn_link>). |
|
|
831 | |
|
|
832 | This also saves round-trips and avoids sending messages to the wrong port |
|
|
833 | (hard to do in Erlang). |
|
|
834 | |
|
|
835 | =back |
|
|
836 | |
|
|
837 | =head1 RATIONALE |
|
|
838 | |
|
|
839 | =over 4 |
|
|
840 | |
|
|
841 | =item Why strings for port and node IDs, why not objects? |
|
|
842 | |
|
|
843 | We considered "objects", but found that the actual number of methods |
|
|
844 | that can be called are quite low. Since port and node IDs travel over |
|
|
845 | the network frequently, the serialising/deserialising would add lots of |
|
|
846 | overhead, as well as having to keep a proxy object everywhere. |
|
|
847 | |
|
|
848 | Strings can easily be printed, easily serialised etc. and need no special |
|
|
849 | procedures to be "valid". |
|
|
850 | |
|
|
851 | And as a result, a miniport consists of a single closure stored in a |
|
|
852 | global hash - it can't become much cheaper. |
|
|
853 | |
|
|
854 | =item Why favour JSON, why not a real serialising format such as Storable? |
|
|
855 | |
|
|
856 | In fact, any AnyEvent::MP node will happily accept Storable as framing |
|
|
857 | format, but currently there is no way to make a node use Storable by |
|
|
858 | default (although all nodes will accept it). |
|
|
859 | |
|
|
860 | The default framing protocol is JSON because a) JSON::XS is many times |
|
|
861 | faster for small messages and b) most importantly, after years of |
|
|
862 | experience we found that object serialisation is causing more problems |
|
|
863 | than it solves: Just like function calls, objects simply do not travel |
|
|
864 | easily over the network, mostly because they will always be a copy, so you |
|
|
865 | always have to re-think your design. |
|
|
866 | |
|
|
867 | Keeping your messages simple, concentrating on data structures rather than |
|
|
868 | objects, will keep your messages clean, tidy and efficient. |
|
|
869 | |
|
|
870 | =back |
52 | |
871 | |
53 | =head1 SEE ALSO |
872 | =head1 SEE ALSO |
|
|
873 | |
|
|
874 | L<AnyEvent::MP::Intro> - a gentle introduction. |
|
|
875 | |
|
|
876 | L<AnyEvent::MP::Kernel> - more, lower-level, stuff. |
|
|
877 | |
|
|
878 | L<AnyEvent::MP::Global> - network maintainance and port groups, to find |
|
|
879 | your applications. |
|
|
880 | |
|
|
881 | L<AnyEvent::MP::LogCatcher> - simple service to display log messages from |
|
|
882 | all nodes. |
54 | |
883 | |
55 | L<AnyEvent>. |
884 | L<AnyEvent>. |
56 | |
885 | |
57 | =head1 AUTHOR |
886 | =head1 AUTHOR |
58 | |
887 | |