Symfony bundle to automatically create and update supervisor configurations for php-amqplib/rabbitmq-bundle (and its predecessor oldsound/rabbitmq-bundle) RabbitMQ consumer daemons.
If you use php-amqplib/rabbitmq-bundle to handle the communication with RabbitMQ, just install supervisor, add this bundle and run
$ app/console rabbitmq-supervisor:rebuildto get a running supervisord instance that automatically manages all your consumer daemons.
When your worker configuration or your code changes, run the command again and all the daemons will be updated.
Install supervisor. e. g. on debian based distributions via apt-get:
$ apt-get install supervisorAdd bundle via composer
$ php composer require phobetor/rabbitmq-supervisor-bundleThis will install the bundle to your project’s vendor directory.
Add the bundle to your project’s AppKernel:
<?php
// app/AppKernel.php
public function registerBundles()
{
$bundles = [
// […]
new Phobetor\RabbitMqSupervisorBundle\RabbitMqSupervisorBundle(),
];
}RabbitMQ supervisor bundle works out of the box with a predefined configuration. If you leave it this way you will end up with this directory structure:
app/supervisor/
└── dev
├── logs
│ ├── stderr.log
│ └── stdout.log
├── supervisord.conf
├── supervisord.log
├── supervisor.pid
├── supervisor.sock
└── worker
├── queue1.conf
├── queue2.conf
├── queue3.conf
└── queue4.confYou can use the following configuration options in your app/config/config.yml:
rabbit_mq_supervisor:
supervisor_instance_identifier: instance_name
paths:
workspace_directory: /path/to/workspace/
configuration_file: /path/to/workspace/supervisord.conf
pid_file: /path/to/workspace/supervisord.pid
sock_file: /path/to/workspace/supervisord.sock
log_file: /path/to/workspace/supervisord.log
worker_configuration_directory: /path/to/workspace/worker/
worker_output_log_file: /path/to/workspace/logs/%kernel.environment%.log
worker_error_log_file: /path/to/workspace/logs/%kernel.environment%.log
commands:
rabbitmq_consumer: user-specific-command:consumer
rabbitmq_multiple_consumer: user-specific-command:multiple-consumer
consumer:
general:
messages: 250 # consumer command option: messages to consume
memory-limit: 32 # consumer command option: allowed memory for this process
debug: true # consumer command option: enable debugging
without-signals: true # consumer command option: disable catching of system signals
worker:
count: 1 # number of workers per consumer
startsecs: 2 # supervisord worker option: seconds to consider program running
autorestart: true # supervisord worker option: if supervisord should restarted program automatically
stopsignal: INT # supervisord worker option: the signal used to kill the program
stopasgroup: true # supervisord worker option: if whole process group should be stopped
stopwaitsecs: 60 # supervisord worker option: seconds to wait after stop signal before sending kill signal
individual:
# override options for specific consumers. you can use the same options for any consumer as in consumer.general
consumer_name_1:
# […]
consumer_name_2:
# […]If you used custom commands before version 2.0, you need to update them. In most case you can just remove everything after the command name.
Build or rebuild the supervisor and worker configuration and start the daemon:
$ app/console rabbitmq-supervisor:rebuildControl the supervisord daemon:
$ app/console rabbitmq-supervisor:control stop
$ app/console rabbitmq-supervisor:control start
$ app/console rabbitmq-supervisor:control restart
$ app/console rabbitmq-supervisor:control hup
