AnyEvent-Task
view release on metacpan or search on metacpan
0.804 2017-02-12
* Bugfix for hung worker timeout not destroying object quickly
0.803 2017-02-12
* Bugfix for timeout errors thrown by checkout with Log::Defer
objects installed.
0.802 2015-10-23
* Stop using AnyEvent::Strict in the tests. It creates an
AE timer at compile time, therefore initializing the event
loop before we've had a chance to fork our server.
* Before forking in fork_task_server, assert that the AE
loop hasn't been initialized.
0.801 2014-02-15
* Bugfix: Fix memory leak of client objects.
* Change: Make hung worker timeout actually terminate the
worker to free up resources immediately.
0.800 2014-02-15
* Major documentation updates
0.750 2013-04-08
* Backwards-incompatible change: The behaviour enabled by the
undocumented client option added in the previous release,
refork_after_error, is now the default behaviour. Instead
there is a new option called dont_refork_after_error to get
back the original behaviour.
* If log_defer_object is passed in when creating a checkout,
operations on this checkout are recorded as Log::Defer
timer events
* Major documentation updates, including working SYNOPSIS
0.720 2013-03-14
* Integration with Log::Defer so workers can log stuff and
it shows up in the client logs
* Complete the implementation of worker error recovery
* Remove undocumented "sk" protocol feature
* Big test-suite refactor/improvements
* Lots more/better docs, including for error handling/logging
Each client maintains a "pool" of connections to worker processes. Every
time a checkout is requested, the request is placed into a first-come,
first-serve queue. Once a worker process becomes available, it is
associated with that checkout until that checkout is garbage collected
which in perl means as soon as it is no longer needed. Each checkout
also maintains a queue of requested method-calls so that as soon as a
worker process is allocated to a checkout, any queued method calls are
filled in order.
"timeout" can be passed as a keyword argument to "checkout". Once a
request is queued up on that checkout, a timer of "timout" seconds
(default is 30, undef means infinity) is started. If the request
completes during this timeframe, the timer is cancelled. If the timer
expires, the worker connection is terminated and an exception is thrown
in the dynamic context of the callback (see the "ERROR HANDLING"
section).
Note that since timeouts are associated with a checkout, checkouts can
be created before the server is started. As long as the server is
running within "timeout" seconds, no error will be thrown and no
requests will be lost. The client will continually try to acquire worker
processes until a server is available, and once one is available it will
attempt to allocate all queued checkouts.
use AnyEvent::Task::Server;
use AnyEvent::Task::Logger;
AnyEvent::Task::Server->new(
name => 'sleeper',
listen => ['unix/', '/tmp/anyevent-task.socket'],
interface => sub {
logger->info('about to compute some operation');
{
my $timer = logger->timer('computing some operation');
select undef,undef,undef, 1; ## sleep for 1 second
}
},
)->run;
Note: Portable server code should never call "sleep" because on some
systems it will interfere with the recoverable worker timeout feature
implemented with "SIGALRM".
In your client code, pass a Log::Defer object in when you create a
'0.023881061050415',
30,
'about to compute some operation'
],
[
'1.025965',
30,
'finished some operation'
]
],
'timers' => {
'computing some operation' => [
'0.024089061050415',
'1.02470206105041'
]
}
};
ERROR HANDLING
In a synchronous program, if you expected some operation to throw an
exception you might wrap it in "eval" like this:
lib/AnyEvent/Task.pm view on Meta::CPAN
=head1 DESIGN
Both client and server are of course built with L<AnyEvent>. However, workers can't use AnyEvent (yet). I've never found a need to do event processing in the worker since if the library you wish to use is already AnyEvent-compatible you can simply us...
Each client maintains a "pool" of connections to worker processes. Every time a checkout is requested, the request is placed into a first-come, first-serve queue. Once a worker process becomes available, it is associated with that checkout until that...
C<timeout> can be passed as a keyword argument to C<checkout>. Once a request is queued up on that checkout, a timer of C<timout> seconds (default is 30, undef means infinity) is started. If the request completes during this timeframe, the timer is c...
Note that since timeouts are associated with a checkout, checkouts can be created before the server is started. As long as the server is running within C<timeout> seconds, no error will be thrown and no requests will be lost. The client will continua...
Because of checkout queuing, the maximum number of worker processes a client will attempt to obtain can be limited with the C<max_workers> argument when creating a client object. If there are more live checkouts than C<max_workers>, the remaining che...
The C<min_workers> argument determines how many "hot-standby" workers should be pre-forked when creating the client. The default is 2 though note that this may change to 0 in the future.
lib/AnyEvent/Task.pm view on Meta::CPAN
use AnyEvent::Task::Server;
use AnyEvent::Task::Logger;
AnyEvent::Task::Server->new(
name => 'sleeper',
listen => ['unix/', '/tmp/anyevent-task.socket'],
interface => sub {
logger->info('about to compute some operation');
{
my $timer = logger->timer('computing some operation');
select undef,undef,undef, 1; ## sleep for 1 second
}
},
)->run;
Note: Portable server code should never call C<sleep> because on some systems it will interfere with the recoverable worker timeout feature implemented with C<SIGALRM>.
In your client code, pass a L<Log::Defer> object in when you create a checkout:
lib/AnyEvent/Task.pm view on Meta::CPAN
'0.023881061050415',
30,
'about to compute some operation'
],
[
'1.025965',
30,
'finished some operation'
]
],
'timers' => {
'computing some operation' => [
'0.024089061050415',
'1.02470206105041'
]
}
};
lib/AnyEvent/Task.pm view on Meta::CPAN
TODO
! max checkout queue size
- start delivering fatal errors to some (at front of queue or back of queue though?)
- write test for this
! docs: write good error handling examples
Make names more consistent between callback::frame backtraces and auto-generated log::defer timers
make server not use AnyEvent so don't have to worry about workers unlinking unix socket in dtors
In a graceful shutdown scenario, servers wait() on all their children before terminating.
- Support relinquishing accept() socket during this period?
Document hung_worker_timeout and SIGALRM stuff
Wire protocol:
- Support something other than JSON? Sereal?
lib/AnyEvent/Task/Client.pm view on Meta::CPAN
my $service = $self->{connect}->[1];
my $worker_guard;
$self->{connecting_workers}->{0 + $worker_guard} = $worker_guard = tcp_connect $host, $service, sub {
my $fh = shift;
delete $self->{connecting_workers}->{0 + $worker_guard};
if (!$fh) {
$self->{total_workers}--;
$self->install_populate_workers_timer;
return;
}
delete $self->{populate_workers_timer};
my $worker; $worker = new AnyEvent::Handle
fh => $fh,
on_read => sub { }, ## So we always have a read watcher and can instantly detect worker deaths
on_error => sub {
my ($worker, $fatal, $message) = @_;
my $checkout = $self->{workers_to_checkouts}->{0 + $worker};
$checkout->{timeout_timer} = undef; ## timer keeps a circular reference
$checkout->throw_fatal_error('worker connection suddenly died') if $checkout;
$self->destroy_worker($worker);
$self->populate_workers;
};
$self->{worker_checkout_counts}->{0 + $worker} = 0;
$self->make_worker_available($worker);
$self->try_to_fill_pending_checkouts;
};
}
}
sub install_populate_workers_timer {
my ($self) = @_;
return if exists $self->{populate_workers_timer};
$self->{populate_workers_timer} = AE::timer 0.2, 1, sub {
$self->populate_workers;
};
}
sub try_to_fill_pending_checkouts {
my ($self) = @_;
return unless @{$self->{pending_checkouts}};
lib/AnyEvent/Task/Client/Checkout.pm view on Meta::CPAN
my %args = (code => $request->[-1]);
$args{name} = $name if defined $name;
$request->[-1] = frame(%args)
unless Callback::Frame::is_frame($request->[-1]);
}
push @{$self->{pending_requests}}, $request;
$self->_install_timeout_timer;
$self->_try_to_fill_requests;
return;
}
sub _install_timeout_timer {
my ($self) = @_;
return if !defined $self->{timeout};
return if exists $self->{timeout_timer};
$self->{timeout_timer} = AE::timer $self->{timeout}, 0, sub {
delete $self->{timeout_timer};
$self->{client}->remove_pending_checkout($self);
if (exists $self->{worker}) {
$self->{client}->destroy_worker($self->{worker});
delete $self->{worker};
}
$self->throw_fatal_error("timed out after $self->{timeout} seconds");
};
lib/AnyEvent/Task/Client/Checkout.pm view on Meta::CPAN
return;
}
my $method_name = $request->[0];
if (!defined $method_name) {
$method_name = '->()';
shift @$request;
}
$self->_install_timeout_timer;
$self->{worker}->push_write( json => [ 'do', {}, @$request, ], );
my $timer;
if ($self->{log_defer_object}) {
$timer = $self->{log_defer_object}->timer($method_name);
}
$self->{cmd_handler} = sub {
my ($handle, $response) = @_;
undef $timer;
my ($response_code, $meta, $response_value) = @$response;
if ($self->{log_defer_object} && $meta->{ld}) {
$self->{log_defer_object}->merge($meta->{ld});
}
if ($response_code eq 'ok') {
local $@ = undef;
$cb->($self, $response_value);
} elsif ($response_code eq 'er') {
$self->_throw_error($response_value);
} else {
die "Unrecognized response_code: $response_code";
}
delete $self->{timeout_timer};
delete $self->{cmd_handler};
$self->_try_to_fill_requests;
};
$self->{worker}->push_read( json => $self->{cmd_handler} );
}
sub DESTROY {
my ($self) = @_;
lib/AnyEvent/Task/Client/Checkout.pm view on Meta::CPAN
if ($self->{fatal_error} || ($self->{error_occurred} && $self->{client} && !$self->{client}->{dont_refork_after_error})) {
$self->{client}->destroy_worker($worker) if $self->{client};
$self->{client}->populate_workers if $self->{client};
} else {
$worker->push_write( json => [ 'dn', {} ] );
$self->{client}->make_worker_available($worker) if $self->{client};
$self->{client}->try_to_fill_pending_checkouts if $self->{client};
}
}
$self->{pending_requests} = $self->{current_cb} = $self->{timeout_timer} = $self->{cmd_handler} = undef;
}
1;
t/error-clears-checkout-queue.t view on Meta::CPAN
);
my $client = AnyEvent::Task::Client->new(
connect => ['unix/', '/tmp/anyevent-task-test.socket'],
);
my $cv = AE::cv;
my $timeout_watcher = AE::timer 0.5, 0, sub {
$cv->send;
};
my $num_exceptions_caught = 0;
frame(code => sub {
my $checkout = $client->checkout;
$checkout->success(sub { ok(1, "first in checkout queue") });
$checkout->success(sub { ok(1, "second in checkout queue") });
use AnyEvent::Util;
use AnyEvent::Task::Server;
use AnyEvent::Task::Client;
use AnyEvent::Task::Logger;
use Test::More tests => 14;
## The point of this test is to verify Log::Defer integration.
## If log_defer_object is passed in when creating a checkout:
## 1) The server can add log messages, timers, data, etc to
## this object by using the AnyEvent::Task::Logger::logger
## 2) Every time a request is placed onto a checkout, a timer
## is started in this object and it is ended once the
## request is fulfilled.
AnyEvent::Task::Server::fork_task_server(
listen => ['unix/', '/tmp/anyevent-task-test.socket'],
interface => { normal =>
sub {
logger->info("hello from", $$);
logger->timer("junk");
1;
},
error =>
sub {
logger->warn("something weird happened");
die "uh oh";
},
sleep =>
sub { select undef,undef,undef,shift; },
},
my $log_defer_object = Log::Defer->new(sub {
my $msg = shift;
is($msg->{logs}->[0]->[2], 'hello from', 'message from client');
is($msg->{logs}->[1]->[2], 'hello from', 'message from worker');
isnt($msg->{logs}->[0]->[3], $msg->{logs}->[1]->[3], 'pids are different');
is($msg->{logs}->[2]->[2], 'after', 'order of msgs ok');
is($msg->{logs}->[3]->[2], 'something weird happened', 'log messages transfered even on error');
is(@{$msg->{timers}}, 5, 'right number of timers');
is($msg->{timers}->[0]->[0], 'normal', 'normal is timer 1');
is($msg->{timers}->[1]->[0], 'junk', 'junk is timer 2');
is($msg->{timers}->[2]->[0], 'sleep', 'sleep is timer 3');
is($msg->{timers}->[3]->[0], 'sleep', 'sleep is timer 4');
is($msg->{timers}->[4]->[0], 'error', 'error is timer 5');
});
$log_defer_object->info("hello from", $$);
$client->checkout(log_defer_object => $log_defer_object)->normal(sub {
my ($checkout, $ret) = @_;
$log_defer_object->info("after");
$checkout->sleep(0.1, sub {});
$cv->recv;
$cv = AE::cv;
$log_defer_object = Log::Defer->new(sub {
my $msg = shift;
is($msg->{timers}->[0]->[0], '->()', "didn't leak first arg when called as code ref");
});
$client->checkout(log_defer_object => $log_defer_object)->('first arg', frame(code => sub {
die "error not thrown by calling interface as a sub?";
}, catch => sub {
ok(1, 'error caught');
$cv->send;
}));
$cv->recv;
t/max_checkouts.t view on Meta::CPAN
connect => ['unix/', '/tmp/anyevent-task-test.socket'],
max_workers => 1,
max_checkouts => 2,
);
my $cv = AE::cv;
my $pid;
my $timeout_watcher = AE::timer 1.0, 0, sub {
print STDERR "hanged, probably because socket was unlinked";
exit;
};
{
$client->checkout->(sub {
my ($checkout, $ret) = @_;
$pid = $ret;
});
t/timeout-log-defer.t view on Meta::CPAN
});
} frame_catch {
$ld->warn("keep alive 2");
my $err = $@;
ok(1, "timeout hit");
ok($err =~ /timed out after/, 'correct err msg');
$error_thrown = 1;
};
}
my $timer = AE::timer 1, 0, sub {
fail("log defer object destroyed");
$cv->send;
};
$cv->recv;
( run in 1.334 second using v1.01-cache-2.11-cpan-3cd7ad12f66 )