Oban.Job (Oban v2.20.1)

A Job is an Ecto schema used for asynchronous execution.

Job changesets are created by your application code and inserted into the database for asynchronous execution. Jobs can be inserted along with other application data as part of a transaction, which guarantees that jobs will only be triggered from a successful transaction.

Summary

Types

args()

changeset()

changeset_fun()

changeset_list()

changeset_list_fun()

errors()

option()

replace_by_state_option()

replace_option()

schedule_in_option()

t()

tags()

time_unit()

unique_field()

unique_option()

unique_period()

unique_state()

unique_state_group()

unique_timestamp()

Functions

format_attempt(job)

Normalize, blame, and format a job's unsaved_error into the stored error format.

new(args, opts \\ [])

Construct a new job changeset ready for insertion into the database.

states()

A canonical list of all possible job states.

to_map(changeset)

Convert a Job changeset into a map suitable for database insertion.

unique_states(atom)

A list of job states tailored to uniqueness constraints.

update(job, changes)

Construct a changeset for updating an existing job with the given changes.

Types

args()

(since 0.1.0)

@type args() :: map()

changeset()

(since 0.1.0)

@type changeset() :: Ecto.Changeset.t(t())

changeset_fun()

(since 0.1.0)

@type changeset_fun() :: (map() -> changeset())

changeset_list()

(since 0.1.0)

@type changeset_list() :: Enumerable.t(changeset())

changeset_list_fun()

(since 0.1.0)

@type changeset_list_fun() :: (map() -> changeset_list())

errors()

(since 0.1.0)

@type errors() :: [%{at: DateTime.t(), attempt: pos_integer(), error: binary()}]

option()

(since 0.1.0)

@type option() ::
  {:args, args()}
  | {:max_attempts, pos_integer()}
  | {:meta, map()}
  | {:priority, 0..9}
  | {:queue, atom() | binary()}
  | {:replace, [replace_option() | replace_by_state_option()]}
  | {:replace_args, boolean()}
  | {:schedule_in, schedule_in_option()}
  | {:scheduled_at, DateTime.t()}
  | {:tags, tags()}
  | {:unique, true | false | nil | [unique_option()]}
  | {:worker, atom() | binary()}

replace_by_state_option()

(since 0.1.0)

@type replace_by_state_option() ::
  {:available, [replace_option()]}
  | {:cancelled, [replace_option()]}
  | {:completed, [replace_option()]}
  | {:discarded, [replace_option()]}
  | {:executing, [replace_option()]}
  | {:retryable, [replace_option()]}
  | {:scheduled, [replace_option()]}

replace_option()

(since 0.1.0)

@type replace_option() :: [
  :args
  | :max_attempts
  | :meta
  | :priority
  | :queue
  | :scheduled_at
  | :tags
  | :worker
]

schedule_in_option()

(since 0.1.0)

@type schedule_in_option() :: pos_integer() | {pos_integer(), time_unit()}

t()

(since 0.1.0)

@type t() :: %Oban.Job{
  __meta__: term(),
  args: args(),
  attempt: non_neg_integer(),
  attempted_at: DateTime.t() | nil,
  attempted_by: [binary()] | nil,
  cancelled_at: DateTime.t() | nil,
  completed_at: DateTime.t() | nil,
  conf: Oban.Config.t() | nil,
  conflict?: boolean(),
  discarded_at: DateTime.t() | nil,
  errors: errors(),
  id: pos_integer(),
  inserted_at: DateTime.t(),
  max_attempts: pos_integer(),
  meta: map(),
  priority: 0..9,
  queue: binary(),
  replace: [replace_option() | replace_by_state_option()] | nil,
  scheduled_at: DateTime.t(),
  state: binary(),
  tags: tags(),
  unique:
    %{
      fields: [unique_field()],
      keys: [atom()],
      period: unique_period(),
      states: unique_state_group() | [unique_state()],
      timestamp: unique_timestamp()
    }
    | nil,
  unsaved_error:
    %{
      kind: Exception.kind(),
      reason: term(),
      stacktrace: Exception.stacktrace()
    }
    | nil,
  worker: binary()
}

tags()

(since 0.1.0)

@type tags() :: [binary()]

time_unit()

(since 0.1.0)

@type time_unit() ::
  :second
  | :seconds
  | :minute
  | :minutes
  | :hour
  | :hours
  | :day
  | :days
  | :week
  | :weeks

unique_field()

(since 0.1.0)

@type unique_field() :: :args | :meta | :queue | :worker

unique_option()

(since 0.1.0)

@type unique_option() ::
  {:fields, [unique_field()]}
  | {:keys, [atom()]}
  | {:period, unique_period()}
  | {:states, [unique_state()]}
  | {:timestamp, unique_timestamp()}

unique_period()

(since 0.1.0)

@type unique_period() :: pos_integer() | {pos_integer(), time_unit()} | :infinity

unique_state()

(since 0.1.0)

@type unique_state() ::
  :available
  | :cancelled
  | :completed
  | :discarded
  | :executing
  | :retryable
  | :scheduled

unique_state_group()

(since 0.1.0)

@type unique_state_group() :: :all | :incomplete | :scheduled | :successful

unique_timestamp()

(since 0.1.0)

@type unique_timestamp() :: :inserted_at | :scheduled_at

Functions

format_attempt(job)

(since 2.14.0)

Normalize, blame, and format a job's unsaved_error into the stored error format.

Formatted errors are stored in a job's errors field.

new(args, opts \\ [])

(since 0.1.0)

@spec new(args(), [option()]) :: changeset()

Construct a new job changeset ready for insertion into the database.

Options

:max_attempts — the maximum number of times a job can be retried if there are errors during execution
:meta — a map containing additional information about the job
:priority — a numerical indicator from 0 to 9 of how important this job is relative to other jobs in the same queue. The lower the number, the higher priority the job.
:queue — a named queue to push the job into. Jobs may be pushed into any queue, regardless of whether jobs are currently being processed for the queue.
:replace - a list of keys to replace per state on a unique conflict
:scheduled_at - a time in the future after which the job should be executed
:schedule_in - the number of seconds until the job should be executed or a tuple containing a number and unit
:tags — a list of tags to group and organize related jobs, i.e. to identify scheduled jobs
:unique — a keyword list of options specifying how uniqueness will be calculated. The options define which fields will be used, for how long, with which keys, and for which states.
:worker — a module to execute the job in. The module must implement the Oban.Worker behaviour.

Examples

Insert a job with the :default queue:

%{id: 1, user_id: 2}
|> Oban.Job.new(queue: :default, worker: MyApp.Worker)
|> Oban.insert()

Generate a pre-configured job for MyApp.Worker:

MyApp.Worker.new(%{id: 1, user_id: 2})

Schedule a job to run in 5 seconds:

MyApp.Worker.new(%{id: 1}, schedule_in: 5)

Schedule a job to run in 5 minutes:

MyApp.Worker.new(%{id: 1}, schedule_in: {5, :minutes})

Insert a job, ensuring that it is unique within the past minute:

MyApp.Worker.new(%{id: 1}, unique: [period: {1, :minute}])

Insert a unique job where the period is compared to the scheduled_at timestamp rather than inserted_at:

MyApp.Worker.new(%{id: 1}, unique: [period: 60, timestamp: :scheduled_at])

Insert a unique job based only on the worker field, and within multiple states:

fields = [:worker]
states = [:available, :scheduled, :executing, :retryable, :completed]

MyApp.Worker.new(%{id: 1}, unique: [fields: fields, period: 60, states: states])

Insert a unique job considering only the worker and specified keys in the args:

keys = [:account_id, :url]
args = %{account_id: 1, url: "https://example.com"}

MyApp.Worker.new(args, unique: [fields: [:args, :worker], keys: keys])

Insert a unique job considering only specified keys in the meta:

unique = [fields: [:meta], keys: [:slug]]

MyApp.Worker.new(%{id: 1}, meta: %{slug: "unique-key"}, unique: unique)

states()

(since 2.1.0)

A canonical list of all possible job states.

This may be used to build up :unique options without duplicating states in application code.

Examples

iex> Oban.Job.states() -- [:completed, :discarded]
~w(scheduled available executing retryable cancelled)a

Job State Transitions

:scheduled—Jobs inserted with scheduled_at in the future are :scheduled. After the scheduled_at time has elapsed Oban's job staging will transition them to :available
:available—Jobs without a future scheduled_at timestamp are inserted as :available and may execute immediately
:executing—Available jobs may be ran, at which point they are :executing
:retryable—Jobs that fail and haven't exceeded their max attempts are transitioned to :retryable and rescheduled until after a backoff period. Once the backoff has elapsed Oban's job staging will transition them back to :available
:completed—Jobs that finish executing succesfully are marked :completed
:cancelled—Jobs that are cancelled intentionally
:discarded—Jobs that fail and exhaust their max attempts, or return a :discard tuple during execution, are marked :discarded

to_map(changeset)

(since 0.9.0)

@spec to_map(Ecto.Changeset.t(t())) :: map()

Convert a Job changeset into a map suitable for database insertion.

Examples

Convert a worker generated changeset into a plain map:

%{id: 123}
|> MyApp.Worker.new()
|> Oban.Job.to_map()

unique_states(atom)

(since 2.20.0)

A list of job states tailored to uniqueness constraints.

Named Groups

:all - All possible job states
:incomplete - States for jobs that haven't completed processing
:scheduled - Only scheduled jobs
:successful - Jobs that aren't cancelled or discarded (default)

Examples

iex> Oban.Job.unique_states(:incomplete)
~w(available scheduled executing retryable)a

iex> Oban.Job.unique_states(:scheduled)
~w(scheduled)a

update(job, changes)

(since 2.20.0)

@spec update(t(), map()) :: Ecto.Changeset.t()

Construct a changeset for updating an existing job with the given changes.

Only a subset of fields are allowed to be updated, and all validations from new/2 are applied to ensure data integrity.

Examples

Update the tags and priority of a job:

Oban.Job.update(job, %{tags: ["urgent"], priority: 5})