All,
I was doing some stress testing for an embedded app that sends lots of
websocket messages, and I can't understand some odd behaviour. Basically mojo
stalls using 100% CPU for many seconds. Doing the following (to simulate heavy
load):
for (1..10000) {
$sc->{websocket}->send({text => $payload});
}
Will send 10000 messages to the client (they all arrive) but then mojo stalls
using 100% CPU for 19 seconds or so. Running nytprof, I get this:
10000 1 1 19.2s 19.2s Mojo::EventEmitter::unsubscribe
50 10000 3.89ms if ($cb) {
51 10000 19.1s $self->{events}{$name} = [grep { $cb ne
$_ } @{$self->{events}{$name}}];
52 10000 9.66ms delete $self->{events}{$name} unless
@{$self->{events}{$name}};
Looking at the above code it seems to suggest that a callback is being used,
but I am not sure where/how it was created? I do use the following paradigm in
my code, could this be the cause of the problem?
my $id = Mojo::IOLoop->client({address => '127.0.0.1', port => 5001} => sub {
my ($loop, $err, $stream) = @_;
...
$stream->on(read => sub {
my ($stream, $bytes) = @_;
# Process input
...
[call a function that would send these 10000 messages to the websocket]
});
});
Thanks,
--
You received this message because you are subscribed to the Google Groups
"Mojolicious" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/mojolicious.
For more options, visit https://groups.google.com/d/optout.
