1 |
=head1 Message Passing for the Non-Blocked Mind |
2 |
|
3 |
=head1 Introduction and Terminology |
4 |
|
5 |
This is a tutorial about how to get the swing of the new L<AnyEvent::MP> |
6 |
module, which allows us to transparently pass messages to our own process |
7 |
and to other processes on another or the same host. |
8 |
|
9 |
What kind of messages? Well, basically a message here means a list of Perl |
10 |
strings, numbers, hashes and arrays, anything that can be expressed as a |
11 |
L<JSON> text (as JSON is used by default in the protocol). |
12 |
|
13 |
It's custom in L<AnyEvent::MP> to have a string which describes the type of the |
14 |
message as first element (this is called a I<tag> in L<AnyEvent::MP>), as some |
15 |
API functions (C<rcv>) support matching it directly. So supposedly you want to |
16 |
send a ping message with your current time to something, this is how such a |
17 |
message might look like (in Perl syntax): |
18 |
|
19 |
['ping', 1251381636] |
20 |
|
21 |
And next you might ask: between which entities are those messages being |
22 |
I<passed>? They are I<passed> between I<ports>. I<ports> are just sources and |
23 |
destinations for messages. How do these ports relate to things you know? Well, |
24 |
each I<port> belongs to a I<node>, and a I<node> is just the UNIX process that |
25 |
runs your L<AnyEvent::MP> application. |
26 |
|
27 |
Each I<node> is distinguished from other I<nodes> running on the same host or |
28 |
multiple hosts in a network by it's I<node ID>. A I<node ID> can be manually |
29 |
assigned or L<AnyEvent::MP> will assign one it self for you. |
30 |
|
31 |
So, you might want to visualize it like this (setup is two nodes: Node C<A> (in |
32 |
UNIX process 7066) with ports C<ABC> and C<DEF> and C<B> (in UNIX process 8321) |
33 |
with ports C<FOO> and C<BAR>). |
34 |
|
35 |
|
36 |
|- PID: 7066 -| |- PID: 8321 -| |
37 |
| | | | |
38 |
| Node ID: A | | Node ID: B | |
39 |
| | | | |
40 |
| Port ABC =|= <----\ /-----> =|= Port FOO | |
41 |
| | X | | |
42 |
| Port DEF =|= <----/ \-----> =|= Port BAR | |
43 |
| | | | |
44 |
|-------------| |-------------| |
45 |
|
46 |
The strings for the ports here are just for illustrative purposes. Even if in |
47 |
reality I<ports> in L<AnyEvent::MP> are also identified by strings they can't |
48 |
be choosen manually and are assigned randomly. These I<port ids> should also |
49 |
not be used directly for other purposes than refering to an endpoint for |
50 |
messages. |
51 |
|
52 |
The next sections will explain the API of L<AnyEvent::MP>. First the API is |
53 |
layed out by simple examples. Later some more complex idioms are introduced, |
54 |
which are maybe useful to solve some realworld purposes. |
55 |
|
56 |
# In this tutorial I'll show you how to write a simple chat server based on |
57 |
# L<AnyEvent::MP>. This example is used because it nicely shows how to organise a |
58 |
# simple application, but keep in mind that every node trusts any other, so this |
59 |
# chat cannot be used to implement a real public chat server and client system, |
60 |
# but it can be used to implement a distributed chat server for example. |
61 |
|
62 |
=head1 Passing Your First Message |
63 |
|
64 |
As start lets have a look at the messaging API. The next example is just a |
65 |
demo to show the basic elements of message passing with L<AnyEvent::MP>. |
66 |
It shout just print: "Ending with: 123". So here the code: |
67 |
|
68 |
use AnyEvent; |
69 |
use AnyEvent::MP; |
70 |
|
71 |
my $end_cv = AnyEvent->condvar; |
72 |
|
73 |
my $port = port; |
74 |
|
75 |
rcv $port, test => sub { |
76 |
my ($data) = @_; |
77 |
$end_cv->send ($data); |
78 |
}; |
79 |
|
80 |
snd $port, test => 123; |
81 |
|
82 |
print "Ending with: " . $end_cv->recv . "\n"; |
83 |
|
84 |
It already contains most functions of the essential L<AnyEvent::MP> API. |
85 |
|
86 |
First there is the C<port> function which will create a I<port> and will return |
87 |
it's I<port id>. |
88 |
|
89 |
That I<port id> can be used to send and receive messages. That I<port id> is a |
90 |
simple string and can be safely passed to other I<nodes> in the network to |
91 |
refer to that specific port (usually used for RPC, where you need to |
92 |
tell the other end which I<port> to send the reply to). |
93 |
|
94 |
Next function is C<rcv>: |
95 |
|
96 |
rcv $port, test => sub { ... }; |
97 |
|
98 |
It sets up a receiver callback on a specific I<port> which needs to be |
99 |
specified as the first argument. The next argument, in this example C<test>, is |
100 |
a I<tag> match. This means that whenever a message, with the first element |
101 |
being the string C<tag>, is received the callback is called with the remaining |
102 |
parts of that message. |
103 |
|
104 |
Messages can be send with the C<snd> function, which looks like this in the |
105 |
example above: |
106 |
|
107 |
snd $port, test => 123; |
108 |
|
109 |
This will send the message C<['test', 123]> to the I<port> with the I<port id> |
110 |
in C<$port>. The receiver got a I<tag> match on C<test> and will call the |
111 |
callback with the first argument being the number C<123>. |
112 |
|
113 |
That callback then just passes that number on to the I<condition variable> |
114 |
C<$end_cv> which will then pass the value to the print. But I<condition |
115 |
variables> are out of the scope of this tutorial. So please consult the |
116 |
L<AnyEvent::Intro> about them. |
117 |
|
118 |
But passing messages inside one process is boring, but before we can continue |
119 |
and take the next step to interprocess message passing we first have to make |
120 |
sure some things have been setup. |
121 |
|
122 |
=head1 System Requirements and System Setup |
123 |
|
124 |
Before we can start with real IPC we have to make sure some things work on your |
125 |
system. |
126 |
|
127 |
First we have to setup a I<shared secret>: for two L<AnyEvent::MP> I<nodes> to |
128 |
be able to communicate with each other and authenticate each other it is |
129 |
necessary to setup the same I<shared secret> for both of them (or use TLS |
130 |
certificates). |
131 |
|
132 |
The easiest way is to set this up is to use the F<aemp> utility: |
133 |
|
134 |
aemp gensecret |
135 |
|
136 |
This creates a F<$HOME/.perl-anyevent-mp> config file and generates a random |
137 |
shared secret. You can copy this file to any other system and then communicate |
138 |
over the network (via TCP) with it. You can also select your own shared secret |
139 |
(F<aemp setsecret>) and for increased security requirements you can even create |
140 |
a TLS certificate (F<aemp gencert>), causing connections to not just be |
141 |
authenticated, but also to be encrypted. |
142 |
|
143 |
Connections will only be successful when the I<nodes> that want to connect to |
144 |
each other have the same I<shared secret> (or successfully verify the TLS |
145 |
certificate of the other side). |
146 |
|
147 |
B<If something does not work as expected, and for example tcpdump shows |
148 |
that the connections are closed almost immediately, you should make sure |
149 |
that F<~/.perl-anyevent-mp> is the same on all hosts/user accounts that |
150 |
you try to connect with each other!> |
151 |
|
152 |
=head1 The Chat Client |
153 |
|
154 |
OK, lets start by implementing the "frontend" of the client. We will |
155 |
develop the client first and postpone the server for later, as the most |
156 |
complex things actually happen in the client. |
157 |
|
158 |
We will use L<AnyEvent::Handle> to do non-blocking IO read on standard |
159 |
input (all of this code deals with actually handling user input, no |
160 |
message passing yet): |
161 |
|
162 |
#!perl |
163 |
|
164 |
use AnyEvent; |
165 |
use AnyEvent::Handle; |
166 |
|
167 |
sub send_message { |
168 |
die "This is where we will send the messages to the server" |
169 |
. "in the next step of this tutorial.\n" |
170 |
} |
171 |
|
172 |
# make an AnyEvent condition variable for the 'quit' condition |
173 |
# (when we want to exit the client). |
174 |
my $quit_cv = AnyEvent->condvar; |
175 |
|
176 |
my $stdin_hdl = AnyEvent::Handle->new ( |
177 |
fh => *STDIN, |
178 |
on_error => sub { $quit_cv->send }, |
179 |
on_read => sub { |
180 |
my ($hdl) = @_; |
181 |
|
182 |
$hdl->push_read (line => sub { |
183 |
my ($hdl, $line) = @_; |
184 |
|
185 |
if ($line =~ /^\/quit/) { # /quit will end the client |
186 |
$quit_cv->send; |
187 |
} else { |
188 |
send_message ($line); |
189 |
} |
190 |
}); |
191 |
} |
192 |
); |
193 |
|
194 |
$quit_cv->recv; |
195 |
|
196 |
This is now a very basic client. Explaining explicitly what |
197 |
L<AnyEvent::Handle> does or what a I<condvar> is all about is out of scope |
198 |
of this document, please consult L<AnyEvent::Intro> or the manual pages |
199 |
for L<AnyEvent> and L<AnyEvent::Handle>. |
200 |
|
201 |
=head1 First Steps Into Messaging |
202 |
|
203 |
To supply the C<send_message> function we now take a look at |
204 |
L<AnyEvent::MP>. This is an example of how it might look like: |
205 |
|
206 |
... # the use lines from the above snippet |
207 |
|
208 |
use AnyEvent::MP; |
209 |
|
210 |
sub send_message { |
211 |
my ($msg) = @_; |
212 |
|
213 |
snd $server_port, message => $msg; |
214 |
} |
215 |
|
216 |
... # the rest of the above script |
217 |
|
218 |
The C<snd> function is exported by L<AnyEvent::MP>, it stands for 'send |
219 |
a message'. The first argument is the I<port> (a I<port> is something |
220 |
that can receive messages, represented by a printable string) of the |
221 |
server which will receive the message. How we get this port will be |
222 |
explained in the next step. |
223 |
|
224 |
The remaining arguments of C<snd> are C<message> and C<$msg>, the first |
225 |
two elements of the I<message> (a I<message> in L<AnyEvent::MP> is a |
226 |
simple list of values, which can be sent to a I<port>). |
227 |
|
228 |
So all the function does is send the two values C<message> (a constant |
229 |
string to tell the server what to expect) and the actual message string. |
230 |
|
231 |
Thats all fine and simple so far, but where do we get the |
232 |
C<$server_port>? Well, we need to get the unique I<port id> of the |
233 |
server's port where it wants to receive all the incoming chat messages. A |
234 |
I<port id> is unfortunately a very unique string, which we are unable to |
235 |
know in advance. But L<AnyEvent::MP> supports the concept of 'registered |
236 |
ports', which is basically a port on the server side registered under |
237 |
a well known name. |
238 |
|
239 |
For example, the server has a port for receiving chat messages with a |
240 |
unique I<port id> and registers it under the name C<chatter>. |
241 |
|
242 |
BTW, these "registered port names" should follow similar rules as Perl |
243 |
identifiers, so you should prefix them with your package/module name to |
244 |
make them unique, unless you use them in the main program. |
245 |
|
246 |
As I<messages> can only be sent to a I<port id> and not just to some name |
247 |
we have to ask the server node for the I<port id> of the port registered |
248 |
as C<chatter>. |
249 |
|
250 |
=head1 Finding The Chatter Port |
251 |
|
252 |
Ok, lots of talk, now some code. Now we will actually get the |
253 |
C<$server_port> from the backend: |
254 |
|
255 |
... |
256 |
|
257 |
use AnyEvent::MP; |
258 |
|
259 |
my $server_node = "127.0.0.1:1299"; |
260 |
|
261 |
my $client_port = port; |
262 |
|
263 |
snd $server_node, lookup => "chatter", $client_port, "resolved"; |
264 |
|
265 |
my $resolved_cv = AnyEvent->condvar; |
266 |
my $server_port; |
267 |
|
268 |
# setup a receiver callback for the 'resolved' message: |
269 |
rcv $client_port, resolved => sub { |
270 |
my ($tag, $chatter_port_id) = @_; |
271 |
|
272 |
print "Resolved the server port 'chatter' to $chatter_port_id\n"; |
273 |
$server_port = $chatter_port_id; |
274 |
|
275 |
$resolved_cv->send; |
276 |
1 |
277 |
}; |
278 |
|
279 |
# lets block the client until we have resolved the server port. |
280 |
$resolved_cv->recv; |
281 |
|
282 |
# now setup another receiver callback for the chat messages: |
283 |
rcv $client_port, message => sub { |
284 |
my ($tag, $msg) = @_; |
285 |
|
286 |
print "chat> $msg\n"; |
287 |
0 |
288 |
}; |
289 |
|
290 |
# send a 'join' message to the server: |
291 |
snd $server_port, join => "$client_port"; |
292 |
|
293 |
sub send_message { ... |
294 |
|
295 |
Now that was a lot of new stuff: |
296 |
|
297 |
First we define the C<$server_node>: In order to refer to another node |
298 |
we need some kind of string to reference it - the node reference. The |
299 |
I<noderef> is basically a comma separated list of C<address:port> |
300 |
pairs. We assume in this tutorial that the server runs on C<127.0.0.1> |
301 |
(localhost) on port 1299, which results in the noderef C<127.0.0.1:1299>. |
302 |
|
303 |
Next, in order to receive a reply from the other node or the server we |
304 |
need to have a I<port> that messages can be sent to. This is what the |
305 |
C<port> function will do for us, it just creates a new local port and |
306 |
returns it's I<port ID> that can then be used to receive messages. |
307 |
|
308 |
When you look carefully, you will see that the first C<snd> uses the |
309 |
C<$server_node> (a noderef) as destination port. Well, what I didn't |
310 |
tell you yet is that each I<node> has a default I<port> to receive |
311 |
messages. The ID of this port is the same as the noderef. |
312 |
|
313 |
This I<default port> provides some special services for us, for example |
314 |
resolving a registered name to a I<port id> (a-ha! finally!). |
315 |
|
316 |
This is exactly what this line does: |
317 |
|
318 |
snd $server_node, lookup => "chatter", $client_port, "resolved"; |
319 |
|
320 |
This sends a message with first element being C<lookup>, followed by the |
321 |
(hopefully) registered port name that we want to resolve to a I<port |
322 |
id>: C<chatter>. And in order for the server node to be able to send us |
323 |
back the resolved I<port ID> we have to tell it where to send it: The |
324 |
result message will be sent to C<$client_port> (the I<port id> of the |
325 |
port we just created), and will have the string C<resolved> as the first |
326 |
element. |
327 |
|
328 |
When the node receives this message, it will look up the name, gobble up |
329 |
all the extra arguments we passed, append the resolved name, and send the |
330 |
resulting list as a message. |
331 |
|
332 |
Next we register a receiver for this C<lookup>-request. |
333 |
|
334 |
rcv $client_port, resolved => sub { |
335 |
my ($tag, $chatter_port_id) = @_; |
336 |
... |
337 |
1 |
338 |
}; |
339 |
|
340 |
This sets up a receiver on our own port for messages with the first |
341 |
element being the string C<resolved>. Receivers can match the contents of |
342 |
the messages before actually executing the specified callback. |
343 |
|
344 |
B<Please note> that the every C<rcv> callback has to return either a true |
345 |
or a false value, indicating whether it is B<successful>/B<done> (true) or |
346 |
still wants to B<continue> (false) receiving messages. |
347 |
|
348 |
In this case we tell the C<$client_port> to look into all the messages |
349 |
it receives and look for the string C<resolved> in the first element of |
350 |
the message. If it is found, the given callback will be called with the |
351 |
message elements as arguments. |
352 |
|
353 |
Using a string as the first element of the message is called I<tagging> |
354 |
the message. It's common practise to code the 'type' of a message into |
355 |
it's first element, as this allows for simple matching. |
356 |
|
357 |
The result message will contain the I<port ID> of the well known port |
358 |
C<chatter> as second element, which will be stored in C<$chatter_port_id>. |
359 |
|
360 |
This port ID will then be stored in C<$server_port>, followed by calling |
361 |
C<send> on $resolved_cv> so the program will continue. |
362 |
|
363 |
The callback then returns a C<1> (a true value), to indicate that it has |
364 |
done it's job and doesn't want to receive further C<resolved> messages. |
365 |
|
366 |
After this the chat message receiver callback is registered with the port: |
367 |
|
368 |
rcv $client_port, message => sub { |
369 |
my ($tag, $msg) = @_; |
370 |
|
371 |
print "chat> $msg\n"; |
372 |
|
373 |
0 |
374 |
}; |
375 |
|
376 |
We assume that all messages that are broadcast to the clients by the |
377 |
server contain the string tag C<message> as first element, and the actual |
378 |
message as second element. The callback returns a false value this time, |
379 |
to indicate that it is not yet done and wants to receive further messages. |
380 |
|
381 |
The last thing to do is to tell the server to send us new chat messages |
382 |
from other clients. We do so by sending the message C<join> followed by |
383 |
our own I<port ID>. |
384 |
|
385 |
# send the server a 'join' message: |
386 |
snd $server_port, join => $client_port; |
387 |
|
388 |
This way the server knows where to send all the new messages to. |
389 |
|
390 |
=head1 The Completed Client |
391 |
|
392 |
This is the complete client script: |
393 |
|
394 |
#!perl |
395 |
|
396 |
use AnyEvent; |
397 |
use AnyEvent::Handle; |
398 |
use AnyEvent::MP; |
399 |
|
400 |
my $server_node = "127.0.0.1:1299"; |
401 |
|
402 |
my $client_port = port; |
403 |
|
404 |
snd $server_node, lookup => "chatter", $client_port, "resolved"; |
405 |
|
406 |
my $resolved_cv = AnyEvent->condvar; |
407 |
my $server_port; |
408 |
|
409 |
# setup a receiver callback for the 'resolved' message: |
410 |
rcv $client_port, resolved => sub { |
411 |
my ($tag, $chatter_port_id) = @_; |
412 |
|
413 |
print "Resolved the server port 'chatter' to $chatter_port_id\n"; |
414 |
$server_port = $chatter_port_id; |
415 |
|
416 |
$resolved_cv->send; |
417 |
1 |
418 |
}; |
419 |
|
420 |
# lets block the client until we have resolved the server port. |
421 |
$resolved_cv->recv; |
422 |
|
423 |
# now setup another receiver callback for the chat messages: |
424 |
rcv $client_port, message => sub { |
425 |
my ($tag, $msg) = @_; |
426 |
|
427 |
print "chat> $msg\n"; |
428 |
0 |
429 |
}; |
430 |
|
431 |
# send a 'join' message to the server: |
432 |
snd $server_port, join => "$client_port"; |
433 |
|
434 |
sub send_message { |
435 |
my ($msg) = @_; |
436 |
|
437 |
snd $server_port, message => $msg; |
438 |
} |
439 |
|
440 |
# make an AnyEvent condition variable for the 'quit' condition |
441 |
# (when we want to exit the client). |
442 |
my $quit_cv = AnyEvent->condvar; |
443 |
|
444 |
my $stdin_hdl = AnyEvent::Handle->new ( |
445 |
fh => *STDIN, |
446 |
on_error => sub { $quit_cv->send }, |
447 |
on_read => sub { |
448 |
my ($hdl) = @_; |
449 |
|
450 |
$hdl->push_read (line => sub { |
451 |
my ($hdl, $line) = @_; |
452 |
|
453 |
if ($line =~ /^\/quit/) { # /quit will end the client |
454 |
$quit_cv->send; |
455 |
} else { |
456 |
send_message ($line); |
457 |
} |
458 |
}); |
459 |
} |
460 |
); |
461 |
|
462 |
$quit_cv->recv; |
463 |
|
464 |
=head1 The Server |
465 |
|
466 |
Ok, we finally come to the server. |
467 |
|
468 |
The server of course also needs to set up a port, and in addition needs to |
469 |
I<register> it, so the clients can find it. |
470 |
|
471 |
Again, let's jump directly into the code: |
472 |
|
473 |
#!perl |
474 |
|
475 |
use AnyEvent; |
476 |
use AnyEvent::MP; |
477 |
|
478 |
become_public "127.0.0.1:1299"; |
479 |
|
480 |
my $chatter_port = port; |
481 |
|
482 |
reg $chatter_port, "chatter"; |
483 |
|
484 |
my %client_ports; |
485 |
|
486 |
rcv $chatter_port, |
487 |
join => sub { |
488 |
my ($tag, $client_port) = @_; |
489 |
|
490 |
print "got new client port: $client_port\n"; |
491 |
$client_ports{$client_port} = 1; |
492 |
|
493 |
0 |
494 |
}, |
495 |
message => sub { |
496 |
my ($tag, $msg) = @_; |
497 |
|
498 |
print "message> $msg\n"; |
499 |
|
500 |
snd $_, message => $msg |
501 |
for keys %client_ports; |
502 |
|
503 |
0 |
504 |
}; |
505 |
|
506 |
AnyEvent->condvar->recv; |
507 |
|
508 |
That is all. Looks much simpler than the client, doesn't it? |
509 |
|
510 |
Let's quickly look over it, as C<rcv> has already been discussed in the |
511 |
client part of this tutorial above. |
512 |
|
513 |
First this: |
514 |
|
515 |
become_public "127.0.0.1:1299"; |
516 |
|
517 |
This will tell our I<node> to become a I<public> node, which means that it |
518 |
can be contacted via TCP. The first argument should be the I<noderef> the |
519 |
server wants to be reachable at. In this case it's the TCP port 1299 on |
520 |
C<127.0.0.1>. |
521 |
|
522 |
Next we set up two receivers, one for the C<join> messages and another one |
523 |
for the actual messages of type C<messsage>. This is done with a single |
524 |
call to C<rcv>, which allows multiple C<< match => $callback >> pairs. |
525 |
|
526 |
In the C<join> callback we receive the client port, which is simply |
527 |
remembered in the C<%client_ports> hash. In the C<message> callback we |
528 |
just iterate through all known C<%client_ports> and relay the message to |
529 |
them. |
530 |
|
531 |
That concludes the server. |
532 |
|
533 |
=head1 The Remaining Problems |
534 |
|
535 |
The implementation as shown still has some bugs. For instance: How does |
536 |
the server know that the client isn't there anymore, so it can clean up |
537 |
the C<%client_ports> hash? Also, the chat messages have no originator, so |
538 |
we don't know who actually sent the message (which would be quite useful |
539 |
for human-to-human interaction: to know who the other one is :). |
540 |
|
541 |
But aside from these issues I hope this tutorial showed you the basics of |
542 |
L<AnyEvent::MP> and explained some common idioms. |
543 |
|
544 |
How to solve the reliability and C<%client_ports> cleanup problem will |
545 |
be explained later in this tutorial (TODO). |
546 |
|
547 |
=head1 Inside The Protocol |
548 |
|
549 |
Now, for the interested parties, let me explain some details about the protocol |
550 |
that L<AnyEvent::MP> nodes use to communicate to each other. If you are not |
551 |
interested you can skip this section. |
552 |
|
553 |
Usually TCP is used for communication. Each I<node>, if configured to be |
554 |
a I<public> node with the C<initialise_node> function will listen on the |
555 |
configured TCP port (default is 4040). |
556 |
|
557 |
If then one I<node> wants to send a message to another I<node> it will |
558 |
connect to the host and port given in the I<port ID>. |
559 |
|
560 |
Then some handshaking occurs to check whether both I<nodes> know the |
561 |
I<shared secret>. Optionally, TLS can be enabled (about how to do this |
562 |
exactly please consult the L<AnyEvent::MP> man page, just a hint: It |
563 |
should be enough to put the private key and (self signed) certificate in |
564 |
the C<~/.aemp-secret> file of all nodes). |
565 |
|
566 |
After the handshake, messages will be exchanged using a serialiser |
567 |
(usually L<JSON> is used for this, but it is also possible to use other |
568 |
serialization formats such as L<Storable>). |
569 |
|
570 |
=head1 SEE ALSO |
571 |
|
572 |
L<AnyEvent> |
573 |
|
574 |
L<AnyEvent::Handle> |
575 |
|
576 |
L<AnyEvent::MP> |
577 |
|
578 |
=head1 AUTHOR |
579 |
|
580 |
Robin Redeker <elmex@ta-sa.org> |
581 |
|