Module: Zizq

Defined in:
lib/zizq/error.rb,
lib/zizq.rb,
lib/zizq/job.rb,
lib/zizq/query.rb,
lib/zizq/client.rb,
lib/zizq/worker.rb,
lib/zizq/backoff.rb,
lib/zizq/crontab.rb,
lib/zizq/version.rb,
lib/zizq/lifecycle.rb,
lib/zizq/resources.rb,
lib/zizq/job_config.rb,
lib/zizq/middleware.rb,
lib/zizq/bulk_enqueue.rb,
lib/zizq/enqueue_with.rb,
lib/zizq/ack_processor.rb,
lib/zizq/configuration.rb,
lib/zizq/crontab_entry.rb,
lib/zizq/resources/job.rb,
lib/zizq/resources/page.rb,
lib/zizq/crontab_builder.rb,
lib/zizq/enqueue_request.rb,
lib/zizq/active_job_config.rb,
lib/zizq/resources/job_page.rb,
lib/zizq/resources/resource.rb,
lib/zizq/resources/cron_entry.rb,
lib/zizq/resources/cron_group.rb,
lib/zizq/resources/error_page.rb,
lib/zizq/crontab_entry_builder.rb,
lib/zizq/resources/error_record.rb,
lib/zizq/resources/job_template.rb,
lib/zizq/resources/error_enumerator.rb

Overview

rbs_inline: enabled frozen_string_literal: true

Defined Under Namespace

Modules: ActiveJobConfig, Job, JobConfig, Middleware, RESET, Resources, UNCHANGED Classes: AckProcessor, Backoff, BulkEnqueue, Client, ClientError, Configuration, ConnectionError, Crontab, CrontabBuilder, CrontabEntry, CrontabEntryBuilder, EnqueueRequest, EnqueueWith, Error, Lifecycle, NotFoundError, Query, ResponseError, ServerError, StreamError, Worker

Constant Summary collapse

VERSION =

: String

"0.3.1"

Class Method Summary collapse

Class Method Details

.build_enqueue_request(job_class, *args, **kwargs) {|req| ... } ⇒ Object

This method is part of a private API. You should avoid using this method if possible, as it may be removed or be changed in the future.

Build an EnqueueRequest for a single job class enqueue.

Yields:

  • (req) 


345
346
347
348
349
350
351
352
353
354
 
# File 'lib/zizq.rb', line 345

def build_enqueue_request(job_class, *args, **kwargs, &block)
  unless job_class.is_a?(Class) && job_class.is_a?(Zizq::JobConfig)
    raise ArgumentError, "#{job_class.inspect} must include Zizq::Job or extend Zizq::ActiveJobConfig"
  end

  zizq_job_class = job_class #: Zizq::JobConfig
  req = zizq_job_class.zizq_enqueue_request(*args, **kwargs)
  yield req if block_given?
  req
end

.client ⇒ Object

Returns a shared client instance built from the global configuration.

The client is memoized so that persistent HTTP connections are reused across calls, reducing TCP connection overhead.



75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
 
# File 'lib/zizq.rb', line 75

def client #: () -> Client
  @client ||= begin
    @client_mutex.synchronize do
      break @client if @client

      configuration.validate!
      @client = Client.new(
        url: configuration.url,
        format: configuration.format,
        ssl_context: configuration.ssl_context,
        read_timeout: configuration.read_timeout,
        stream_idle_timeout: configuration.stream_idle_timeout
      )
    end
  end
end

.configuration ⇒ Object

Returns the client configuration.

The configuration can be updated by calling [‘Zizq::configure`].

This configuration is for the client only. Worker parameters are configured on a per-run basis for flexibility.



50
51
52
 
# File 'lib/zizq.rb', line 50

def configuration #: () -> Configuration
  @configuration ||= Configuration.new
end

.configure ⇒ Object

Yields the global configuration ready for updates, which should be done during application initialization, before any jobs are enqueued or worked.

Zizq.configure do |c|
  c.url = "http://localhost:7890"
  c.format = :msgpack
  c.dequeue_middleware.use(MyDequeueMiddleware.new)
end



63
64
65
66
67
68
69
 
# File 'lib/zizq.rb', line 63

def configure #: () { (Configuration) -> void } -> void
  yield configuration
ensure
  # shared client is potentially stale
  @client&.close
  @client = nil
end

.crontab(name) ⇒ Object

Obtain a handle for the given Crontab schedule.

This is a lazy operation. The schedule data is only fetched from the Zizq server upon first accessing data within the schedule.

Zizq.crontab("default").paused?
Zizq.crontab("default").resume!



203
204
205
 
# File 'lib/zizq.rb', line 203

def crontab(name)
  Crontab.new(name)
end

.crontabs ⇒ Object

Return a list of all available Crontab schedules.



126
127
128
 
# File 'lib/zizq.rb', line 126

def crontabs #: () -> Array[String]
  Zizq.client.list_cron_groups
end

.define_crontab(name, timezone: nil, paused: nil, &block) ⇒ Object

Define (or redefine) a Crontab schedule.

This requires a Pro license on the Zizq server.

Crontabs are used to define collections of recurring jobs that run on a specified schedule, such as at 2am on every Monday. Each entry on the Crontab is a single job enqueue, which the Zizq server automatically triggers at the correct point in time. Zizq uses standard Cron expression syntax (with support for seconds via 6-fields) to define entries.

This is designed to be idempotent. You can define a schedule somewhere in your application startup process (after ‘Zizq.configure`) and it doesn’t matter if multiple process all define the same schedule. Zizq is smart enough to handle this correctly.

Entire schedules, and individual entries on a schedule, can be paused and resumed.

By default schedules operate in the system time zone of the Zizq server but an explicit IANA timezone name can be specified when defining the Crontab.

This method sends exactly one request to the Zizq server upon completion of the block. Any existing entries are retained. Any new entries are added, any absent entries are removed, and any modified entries are replaced. In short, whatever the block defines is what the entire resulting Crontab schedule will look like.

Zizq.define_crontab("example", timezone: "Europe/Rome") do |cron|
  cron.define_entry(
    "refresh_data_warehose",
    "*/15 * * * *"
  ).enqueue(RefreshDataWarehoseJob, incremental: true)

  cron.define_entry(
    "expire_acess_tokens",
    "*/10 * * * * *"
  ).enqueue_raw(
    queue: "identity-server",
    type: "expire_access_tokens",
    priority: 100,
    payload: {},
  )
end

When jobs are pushed to the queue at their execution time, Zizq handles this atomically, so there is no risk of a duplicate enqueue for the same schedule tick. However, if you have long-running jobs that should not be permitted to overlap, such as in the case your schedule runs every 10 seconds but jobs can take 30 seconds to execute, you should consider using unique jobs.



187
188
189
190
191
 
# File 'lib/zizq.rb', line 187

def define_crontab(name, timezone: nil, paused: nil, &block)
  crontab = Crontab.new(name)
  crontab.redefine(timezone:, paused:, &block)
  crontab
end

.enqueue(job_class, *args, **kwargs, &block) ⇒ Object

Enqueue a job by class with positional and keyword arguments.

By default all arguments are serialized as JSON, which means hashes with symbol keys will become hashes with string keys. The serialization behaviour can be changed by implementing ‘::zizq_serialize` and `::zizq_deserialize` as class methods on the job class.

Default job options can be overridden at enqueue-time by providing a block which receives a mutable ‘Zizq::EnqueueRequest` instance.

Zizq.enqueue(SendEmailJob, 42, template: "welcome")
Zizq.enqueue(SendEmailJob, 42) { |o| o.queue = "priority" }

Job classes may also override ‘::zizq_enqueue_options` to implement dynamically computed options, such as dynamic prioritisation. This class method accepts the same arguments as the `#perform` method and returns an instance of `Zizq::EnqueueRequest`. Any overrides may call `super` and modify the result.

class SendEmailJob
  include Zizq::Job

  zizq_priority 1000

  def self.zizq_enqueue_options(user_id, template:)
    opts = super
    opts.priority /= 2 if template == "welcome"
    opts
  end

  def perform(user_id, template:)
    # ...
  end
end



247
248
249
250
251
 
# File 'lib/zizq.rb', line 247

def enqueue(job_class, *args, **kwargs, &block)
  req = build_enqueue_request(job_class, *args, **kwargs, &block)
  req = configuration.enqueue_middleware.call(req)
  client.enqueue(**req.to_enqueue_params)
end

.enqueue_bulk {|builder| ... } ⇒ Object

Yields:

  • (builder) 


325
326
327
328
329
330
331
332
333
334
335
 
# File 'lib/zizq.rb', line 325

def enqueue_bulk(&block)
  builder = BulkEnqueue.new
  yield builder
  return [] if builder.requests.empty?

  jobs = builder.requests.map do |req|
    configuration.enqueue_middleware.call(req).to_enqueue_params
  end

  client.enqueue_bulk(jobs:)
end

.enqueue_raw(queue:, type:, payload:, **opts) ⇒ Object

Enqueue a job by providing raw inputs to the Zizq server.

This is for advanced use cases such as enqueueing jobs for consumption in other programming languages.

Zizq.enqueue_raw(
  queue: "emails",
  type: "send_email",
  payload: {user_id: 42, template: "welcome"}
)

If using this method to enqueue a job that is intended for consumption in the Ruby client itself a custom dispatcher implementation is likely required:

Zizq.configure do |c|
  c.dispatcher = MyDispatcher.new
end



283
284
285
286
287
 
# File 'lib/zizq.rb', line 283

def enqueue_raw(queue:, type:, payload:, **opts)
  req = EnqueueRequest.new(queue:, type:, payload:, **opts)
  req = configuration.enqueue_middleware.call(req)
  client.enqueue(**req.to_enqueue_params)
end

.enqueue_with(**overrides) ⇒ Object

Enqueue multiple jobs atomically in a single bulk request.

This can significantly imprive throughput when many jobs need to be enqueued collectively. There is no upper limit on the number of jobs in the request though generally it is probably wise to keep this to less than 1000 jobs unless you have strong atomicity requuirements for a larger number of jobs..

Yields a builder object whose ‘#enqueue` method accepts the same arguments as `Zizq.enqueue`. All collected jobs are sent as a single `POST /jobs/bulk` request and an array of jobs is returned in the same order as the inputs.

Zizq.enqueue_bulk do |b|
  b.enqueue(ProcessPaymentJob, 7)
  b.enqueue(SendEmailJob, 42, template: "welcome")
  b.enqueue(SendEmailJob, 42) { |o| o.queue = "priority" }
end

Build a scoped enqueue helper that applies the given option overrides to every enqueue routed through it. Equivalent to using the block form of ‘Zizq.enqueue`, but composable and reusable.

Zizq.enqueue_with(ready_at: Time.now + 3600).enqueue(MyJob, 42)
Zizq.enqueue_with(priority: 0).enqueue_bulk { |b| ... }

See ‘Zizq::EnqueueWith` for details.



319
320
321
 
# File 'lib/zizq.rb', line 319

def enqueue_with(**overrides)
  EnqueueWith.new(self, overrides)
end

.query ⇒ Object

Start a query to retrieve or modify job data.



121
122
123
 
# File 'lib/zizq.rb', line 121

def query(...)
  Query.new(...)
end

.queues ⇒ Object

List all distinct queue names on the server.



106
107
108
 
# File 'lib/zizq.rb', line 106

def queues #: () -> Array[String]
  client.get_queues
end

.reset! ⇒ Object

Resets all global state: configuration and shared client. Intended for use in tests.



94
95
96
97
98
 
# File 'lib/zizq.rb', line 94

def reset! #: () -> void
  @client&.close
  @client = nil
  @configuration = nil
end

.server_version ⇒ Object

Server version string.



101
102
103
 
# File 'lib/zizq.rb', line 101

def server_version #: () -> String
  client.server_version
end