Queue

Job

class connector.queue.job.Job(func=None, model_name=None, args=None, kwargs=None, priority=None, eta=None, job_uuid=None, max_retries=None, description=None, channel=None, identity_key=None)

Bases: object

A Job is a task to execute.

uuid

Id (UUID) of the job.

state

State of the job, can pending, enqueued, started, done or failed. The start state is pending and the final state is done.

retry

The current try, starts at 0 and each time the job is executed, it increases by 1.

max_retries

The maximum number of retries allowed before the job is considered as failed.

func_name

Name of the function (in the form module.function_name).

args

Arguments passed to the function when executed.

kwargs

Keyword arguments passed to the function when executed.

func_string

Full string representing the function to be executed, ie. module.function(args, kwargs)

description

Human description of the job.

func

The python function itself.

model_name

OpenERP model on which the job will run.

priority

Priority of the job, 0 being the higher priority.

date_created

Date and time when the job was created.

date_enqueued

Date and time when the job was enqueued.

date_started

Date and time when the job was started.

date_done

Date and time when the job was done.

result

A description of the result (for humans).

exc_info

Exception information (traceback) when the job failed.

user_id

OpenERP user id which created the job

eta

Estimated Time of Arrival of the job. It will not be executed before this date/time.

channel

The complete name of the channel to use to process the job. If provided it overrides the one defined on the job’s function.

description

eta

func

func_string

identity_key

perform(session)

Execute the job.

The job is executed with the user which has initiated it.

Parameters:session (ConnectorSession) – session to execute the job

postpone(result=None, seconds=None)

Write an estimated time arrival to n seconds later than now. Used when an retryable exception want to retry a job later.

related_action(session)

set_done(result=None)

set_enqueued()

set_failed(exc_info=None)

set_pending(result=None, reset_retry=True)

set_started()

uuid

Job ID, this is an UUID

class connector.queue.job.JobStorage

Bases: object

Interface for the storage of jobs

exists(job_uuid)

Returns if a job still exists in the storage.

load(job_uuid)

Read the job’s data from the storage

store(job_)

Store a job

class connector.queue.job.OpenERPJobStorage(session)

Bases: connector.queue.job.JobStorage

Store a job on OpenERP

db_record(job_)

db_record_from_uuid(job_uuid)

enqueue(func, model_name=None, args=None, kwargs=None, priority=None, eta=None, max_retries=None, description=None, channel=None, identity_key=None)

Create a Job and enqueue it in the queue. Return the job uuid.

This expects the arguments specific to the job to be already extracted from the ones to pass to the job function.

enqueue_resolve_args(func, *args, **kwargs)

Create a Job and enqueue it in the queue. Return the job uuid.

exists(job_uuid)

Returns if a job still exists in the storage.

job_record_with_same_identity_key(identity_key)

Check if a job to be executed with the same key exists.

load(job_uuid)

Read a job from the Database

store(job_)

Store the Job

connector.queue.job.identity_exact(job_)

Identity function using the model, method and all arguments as key When used, this identity key will have the effect that when a job should be created and a pending job with the exact same recordset and arguments, the second will not be created. It should be used with the identity_key argument: .. python:

from odoo.addons.connector.queue.job import identity_exact
# [...]
    my_delayable_export_record_method.delay(
        session, "my.model", self.id, force=True,
        identity_key=identity_exact)

Alternative identity keys can be built using the various fields of the job. For example, you could compute a hash using only some arguments of the job. .. python:

def identity_example(job_):
    hasher = hashlib.sha1()
    hasher.update(str(job_.args))
    hasher.update(str(job_.kwargs.get('foo', '')))
    return hasher.hexdigest()

Usually you will probably always want to include at least the name of the model and method.

connector.queue.job.job(func=None, default_channel='root', retry_pattern=None)

Decorator for jobs.

Optional argument:

Parameters:

• default_channel – the channel wherein the job will be assigned. This channel is set at the installation of the module and can be manually changed later using the views.
• retry_pattern (dict(retry_count,retry_eta_seconds)) – The retry pattern to use for postponing a job. If a job is postponed and there is no eta specified, the eta will be determined from the dict in retry_pattern. When no retry pattern is provided, jobs will be retried after RETRY_INTERVAL seconds.

Add a delay attribute on the decorated function.

When delay is called, the function is transformed to a job and stored in the OpenERP queue.job model. The arguments and keyword arguments given in delay will be the arguments used by the decorated function when it is executed.

retry_pattern is a dict where keys are the count of retries and the values are the delay to postpone a job.

The delay() function of a job takes the following arguments:

session

Current ConnectorSession

model_name

name of the model on which the job has something to do

*args and **kargs

Arguments and keyword arguments which will be given to the called function once the job is executed. They should be pickle-able.

There are 5 special and reserved keyword arguments that you can use:

• priority: priority of the job, the smaller is the higher priority.

  Default is 10.

• max_retries: maximum number of retries before giving up and set

  the job state to ‘failed’. A value of 0 means infinite retries. Default is 5.

• eta: the job can be executed only after this datetime

  (or now + timedelta if a timedelta or integer is given)

• description : a human description of the job,

  intended to discriminate job instances (Default is the func.__doc__ or

  ‘Function %s’ % func.__name__)

• channel : The complete name of the channel to use to process the job. If

  provided it overrides the one defined on the job’s function.

Example:

@job
def export_one_thing(session, model_name, one_thing):
    # work
    # export one_thing

export_one_thing(session, 'a.model', the_thing_to_export)
# => normal and synchronous function call

export_one_thing.delay(session, 'a.model', the_thing_to_export)
# => the job will be executed as soon as possible

export_one_thing.delay(session, 'a.model', the_thing_to_export,
                       priority=30, eta=60*60*5)
# => the job will be executed with a low priority and not before a
# delay of 5 hours from now

@job(default_channel='root.subchannel')
def export_one_thing(session, model_name, one_thing):
    # work
    # export one_thing

@job(retry_pattern={1: 10 * 60,
                    5: 20 * 60,
                    10: 30 * 60,
                    15: 12 * 60 * 60})
def retryable_example(session):
    # 5 first retries postponed 10 minutes later
    # retries 5 to 10 postponed 20 minutes later
    # retries 10 to 15 postponed 30 minutes later
    # all subsequent retries postponed 12 hours later
    raise RetryableJobError('Must be retried later')

retryable_example.delay(session)

See also: related_action() a related action can be attached to a job

connector.queue.job.related_action(action=<function <lambda>>, **kwargs)

Attach a Related Action to a job.

A Related Action will appear as a button on the OpenERP view. The button will execute the action, usually it will open the form view of the record related to the job.

The action must be a callable that responds to arguments:

session, job, **kwargs

Example usage:

def related_action_partner(session, job):
    model = job.args[0]
    partner_id = job.args[1]
    # eventually get the real ID if partner_id is a binding ID
    action = {
        'name': _("Partner"),
        'type': 'ir.actions.act_window',
        'res_model': model,
        'view_type': 'form',
        'view_mode': 'form',
        'res_id': partner_id,
    }
    return action

@job
@related_action(action=related_action_partner)
def export_partner(session, model_name, partner_id):
    # ...

The kwargs are transmitted to the action:

def related_action_product(session, job, extra_arg=1):
    assert extra_arg == 2
    model = job.args[0]
    product_id = job.args[1]

@job
@related_action(action=related_action_product, extra_arg=2)
def export_product(session, model_name, product_id):
    # ...

connector.queue.job.whitelist_unpickle_global(fn_or_class)

Allow a function or class to be used in jobs

By default, the only types allowed to be used in job arguments are:

• the builtins: str/unicode, int/long, float, bool, tuple, list, dict, None
• the pre-registered: datetime.datetime datetime.timedelta

If you need to use an argument in a job which is not in this whitelist, you can add it by using:

whitelist_unpickle_global(fn_or_class_to_register)

Worker

Queue

Models