Sidekiq Style Guide

This document outlines various guidelines that should be followed when adding or modifying Sidekiq workers.

ApplicationWorker

All workers should include ApplicationWorker instead of Sidekiq::Worker, which adds some convenience methods and automatically sets the queue based on the worker's name.

Dedicated Queues

All workers should use their own queue, which is automatically set based on the worker class name. For a worker named ProcessSomethingWorker, the queue name would be process_something. If you're not sure what queue a worker uses, you can find it using SomeWorker.queue. There is almost never a reason to manually override the queue name using sidekiq_options queue: :some_queue.

After adding a new queue, run bin/rake gitlab:sidekiq:all_queues_yml:generate to regenerate app/workers/all_queues.yml or ee/app/workers/all_queues.yml so that it can be picked up by sidekiq-cluster. Additionally, run bin/rake gitlab:sidekiq:sidekiq_queues_yml:generate to regenerate config/sidekiq_queues.yml.

Queue Namespaces

While different workers cannot share a queue, they can share a queue namespace.

Defining a queue namespace for a worker makes it possible to start a Sidekiq process that automatically handles jobs for all workers in that namespace, without needing to explicitly list all their queue names. If, for example, all workers that are managed by sidekiq-cron use the cronjob queue namespace, we can spin up a Sidekiq process specifically for these kinds of scheduled jobs. If a new worker using the cronjob namespace is added later on, the Sidekiq process also picks up jobs for that worker (after having been restarted), without the need to change any configuration.

A queue namespace can be set using the queue_namespace DSL class method:

class SomeScheduledTaskWorker
  include ApplicationWorker

  queue_namespace :cronjob

  # ...
end

Behind the scenes, this sets SomeScheduledTaskWorker.queue to cronjob:some_scheduled_task. Commonly used namespaces have their own concern module that can easily be included into the worker class, and that may set other Sidekiq options besides the queue namespace. CronjobQueue, for example, sets the namespace, but also disables retries.

bundle exec sidekiq is namespace-aware, and listens on all queues in a namespace (technically: all queues prefixed with the namespace name) when a namespace is provided instead of a simple queue name in the --queue (-q) option, or in the :queues: section in config/sidekiq_queues.yml.

Note that adding a worker to an existing namespace should be done with care, as the extra jobs take resources away from jobs from workers that were already there, if the resources available to the Sidekiq process handling the namespace are not adjusted appropriately.

Versioning

Version can be specified on each Sidekiq worker class. This is then sent along when the job is created.

class FooWorker
  include ApplicationWorker

  version 2

  def perform(*args)
    if job_version == 2
      foo = args.first['foo']
    else
      foo = args.first
    end
  end
end

Under this schema, any worker is expected to be able to handle any job that was enqueued by an older version of that worker. This means that when changing the arguments a worker takes, you must increment the version (or set version 1 if this is the first time a worker's arguments are changing), but also make sure that the worker is still able to handle jobs that were queued with any earlier version of the arguments. From the worker's perform method, you can read self.job_version if you want to specifically branch on job version, or you can read the number or type of provided arguments.

Idempotent Jobs

It's known that a job can fail for multiple reasons. For example, network outages or bugs. In order to address this, Sidekiq has a built-in retry mechanism that is used by default by most workers within GitLab.

It's expected that a job can run again after a failure without major side-effects for the application or users, which is why Sidekiq encourages jobs to be idempotent and transactional.

As a general rule, a worker can be considered idempotent if:

It can safely run multiple times with the same arguments.
Application side-effects are expected to happen only once (or side-effects of a second run do not have an effect).

A good example of that would be a cache expiration worker.

A job scheduled for an idempotent worker is deduplicated when an unstarted job with the same arguments is already in the queue.

Ensuring a worker is idempotent

Make sure the worker tests pass using the following shared example:

include_examples 'an idempotent worker' do
  it 'marks the MR as merged' do
    # Using subject inside this block will process the job multiple times
    subject

    expect(merge_request.state).to eq('merged')
  end
end

Use the perform_multiple method directly instead of job.perform (this helper method is automatically included for workers).

Declaring a worker as idempotent

class IdempotentWorker
  include ApplicationWorker

  # Declares a worker is idempotent and can
  # safely run multiple times.
  idempotent!

  # ...
end

It's encouraged to only have the idempotent! call in the top-most worker class, even if the perform method is defined in another class or module.

If the worker class isn't marked as idempotent, a cop fails. Consider skipping the cop if you're not confident your job can safely run multiple times.

Deduplication

When a job for an idempotent worker is enqueued while another unstarted job is already in the queue, GitLab drops the second job. The work is skipped because the same work would be done by the job that was scheduled first; by the time the second job executed, the first job would do nothing.

Strategies

GitLab supports two deduplication strategies:

until_executing
until_executed

More deduplication strategies have been suggested. If you are implementing a worker that could benefit from a different strategy, please comment in the issue.

Until Executing

This strategy takes a lock when a job is added to the queue, and removes that lock before the job starts.

For example, AuthorizedProjectsWorker takes a user ID. When the worker runs, it recalculates a user's authorizations. GitLab schedules this job each time an action potentially changes a user's authorizations. If the same user is added to two projects at the same time, the second job can be skipped if the first job hasn't begun, because when the first job runs, it creates the authorizations for both projects.

module AuthorizedProjectUpdate
  class UserRefreshOverUserRangeWorker
    include ApplicationWorker

    deduplicate :until_executing
    idempotent!

    # ...
  end
end

Until Executed

This strategy takes a lock when a job is added to the queue, and removes that lock after the job finishes. It can be used to prevent jobs from running simultaneously multiple times.

module Ci
  class BuildTraceChunkFlushWorker
    include ApplicationWorker

    deduplicate :until_executed
    idempotent!

    # ...
  end
end

Scheduling jobs in the future

GitLab doesn't skip jobs scheduled in the future, as we assume that the state has changed by the time the job is scheduled to execute. Deduplication of jobs scheduled in the feature is possible for both until_executed and until_executing strategies.

If you do want to deduplicate jobs scheduled in the future, this can be specified on the worker by passing including_scheduled: true argument when defining deduplication strategy:

module AuthorizedProjectUpdate
  class UserRefreshOverUserRangeWorker
    include ApplicationWorker

    deduplicate :until_executing, including_scheduled: true
    idempotent!

    # ...
  end
end

Limited capacity worker

It is possible to limit the number of concurrent running jobs for a worker class by using the LimitedCapacity::Worker concern.

The worker must implement three methods:

perform_work: The concern implements the usual perform method and calls perform_work if there's any available capacity.
remaining_work_count: Number of jobs that have work to perform.
max_running_jobs: Maximum number of jobs allowed to run concurrently.

class MyDummyWorker
  include ApplicationWorker
  include LimitedCapacity::Worker

  def perform_work(*args)
  end

  def remaining_work_count(*args)
    5
  end

  def max_running_jobs
    25
  end
end

Additional to the regular worker, a cron worker must be defined as well to backfill the queue with jobs. the arguments passed to perform_with_capacity are passed to the perform_work method.

class ScheduleMyDummyCronWorker
  include ApplicationWorker
  include CronjobQueue

  def perform(*args)
    MyDummyWorker.perform_with_capacity(*args)
  end
end

How many jobs are running?

It runs max_running_jobs at almost all times.

The cron worker checks the remaining capacity on each execution and it schedules at most max_running_jobs jobs. Those jobs on completion re-enqueue themselves immediately, but not on failure. The cron worker is in charge of replacing those failed jobs.

Handling errors and idempotence

This concern disables Sidekiq retries, logs the errors, and sends the job to the dead queue. This is done to have only one source that produces jobs and because the retry would occupy a slot with a job to perform in the distant future.

We let the cron worker enqueue new jobs, this could be seen as our retry and back off mechanism because the job might fail again if executed immediately. This means that for every failed job, we run at a lower capacity until the cron worker fills the capacity again. If it is important for the worker not to get a backlog, exceptions must be handled in #perform_work and the job should not raise.

The jobs are deduplicated using the :none strategy, but the worker is not marked as idempotent!.

Metrics

This concern exposes three Prometheus metrics of gauge type with the worker class name as label:

limited_capacity_worker_running_jobs
limited_capacity_worker_max_running_jobs
limited_capacity_worker_remaining_work_count

Job urgency

Jobs can have an urgency attribute set, which can be :high, :low, or :throttled. These have the below targets:

Urgency	Queue Scheduling Target	Execution Latency Requirement
`:high`	10 seconds	p50 of 1 second, p99 of 10 seconds
`:low`	1 minute	Maximum run time of 5 minutes
`:throttled`	None	Maximum run time of 5 minutes

To set a job's urgency, use the urgency class method:

class HighUrgencyWorker
  include ApplicationWorker

  urgency :high

  # ...
end

Latency sensitive jobs

If a large number of background jobs get scheduled at once, queueing of jobs may occur while jobs wait for a worker node to be become available. This is normal and gives the system resilience by allowing it to gracefully handle spikes in traffic. Some jobs, however, are more sensitive to latency than others. Examples of these jobs include:

A job which updates a merge request following a push to a branch.
A job which invalidates a cache of known branches for a project after a push to the branch.
A job which recalculates the groups and projects a user can see after a change in permissions.
A job which updates the status of a CI pipeline after a state change to a job in the pipeline.

When these jobs are delayed, the user may perceive the delay as a bug: for example, they may push a branch and then attempt to create a merge request for that branch, but be told in the UI that the branch does not exist. We deem these jobs to be urgency :high.

Extra effort is made to ensure that these jobs are started within a very short period of time after being scheduled. However, in order to ensure throughput, these jobs also have very strict execution duration requirements:

The median job execution time should be less than 1 second.
99% of jobs should complete within 10 seconds.

If a worker cannot meet these expectations, then it cannot be treated as a urgency :high worker: consider redesigning the worker, or splitting the work between two different workers, one with urgency :high code that executes quickly, and the other with urgency :low, which has no execution latency requirements (but also has lower scheduling targets).

Changing a queue's urgency

On GitLab.com, we run Sidekiq in several shards, each of which represents a particular type of workload.

When changing a queue's urgency, or adding a new queue, we need to take into account the expected workload on the new shard. Note that, if we're changing an existing queue, there is also an effect on the old shard, but that always reduces work.

To do this, we want to calculate the expected increase in total execution time and RPS (throughput) for the new shard. We can get these values from:

The Queue Detail dashboard has values for the queue itself. For a new queue, we can look for queues that have similar patterns or are scheduled in similar circumstances.
The Shard Detail dashboard has Total Execution Time and Throughput (RPS). The Shard Utilization panel displays if there is currently any excess capacity for this shard.

We can then calculate the RPS * average runtime (estimated for new jobs) for the queue we're changing to see what the relative increase in RPS and execution time we expect for the new shard:

new_queue_consumption = queue_rps * queue_duration_avg
shard_consumption = shard_rps * shard_duration_avg

(new_queue_consumption / shard_consumption) * 100

If we expect an increase of less than 5%, then no further action is needed.

Otherwise, please ping @gitlab-org/scalability on the merge request and ask for a review.

Jobs with External Dependencies

Most background jobs in the GitLab application communicate with other GitLab services. For example, PostgreSQL, Redis, Gitaly, and Object Storage. These are considered to be "internal" dependencies for a job.

However, some jobs are dependent on external services in order to complete successfully. Some examples include:

Jobs which call web-hooks configured by a user.
Jobs which deploy an application to a k8s cluster configured by a user.

These jobs have "external dependencies". This is important for the operation of the background processing cluster in several ways:

Most external dependencies (such as web-hooks) do not provide SLOs, and therefore we cannot guarantee the execution latencies on these jobs. Since we cannot guarantee execution latency, we cannot ensure throughput and therefore, in high-traffic environments, we need to ensure that jobs with external dependencies are separated from high urgency jobs, to ensure throughput on those queues.
Errors in jobs with external dependencies have higher alerting thresholds as there is a likelihood that the cause of the error is external.

class ExternalDependencyWorker
  include ApplicationWorker

  # Declares that this worker depends on
  # third-party, external services in order
  # to complete successfully
  worker_has_external_dependencies!

  # ...
end

A job cannot be both high urgency and have external dependencies.

CPU-bound and Memory-bound Workers

Workers that are constrained by CPU or memory resource limitations should be annotated with the worker_resource_boundary method.

Most workers tend to spend most of their time blocked, waiting on network responses from other services such as Redis, PostgreSQL, and Gitaly. Since Sidekiq is a multi-threaded environment, these jobs can be scheduled with high concurrency.

Some workers, however, spend large amounts of time on-CPU running logic in Ruby. Ruby MRI does not support true multi-threading - it relies on the GIL to greatly simplify application development by only allowing one section of Ruby code in a process to run at a time, no matter how many cores the machine hosting the process has. For IO bound workers, this is not a problem, since most of the threads are blocked in underlying libraries (which are outside of the GIL).

If many threads are attempting to run Ruby code simultaneously, this leads to contention on the GIL which has the effect of slowing down all processes.

In high-traffic environments, knowing that a worker is CPU-bound allows us to run it on a different fleet with lower concurrency. This ensures optimal performance.

Likewise, if a worker uses large amounts of memory, we can run these on a bespoke low concurrency, high memory fleet.

Note that memory-bound workers create heavy GC workloads, with pauses of 10-50ms. This has an impact on the latency requirements for the worker. For this reason, memory bound, urgency :high jobs are not permitted and fail CI. In general, memory bound workers are discouraged, and alternative approaches to processing the work should be considered.

If a worker needs large amounts of both memory and CPU time, it should be marked as memory-bound, due to the above restriction on high urgency memory-bound workers.

Declaring a Job as CPU-bound

This example shows how to declare a job as being CPU-bound.

class CPUIntensiveWorker
  include ApplicationWorker

  # Declares that this worker will perform a lot of
  # calculations on-CPU.
  worker_resource_boundary :cpu

  # ...
end

Determining whether a worker is CPU-bound

We use the following approach to determine whether a worker is CPU-bound:

In the Sidekiq structured JSON logs, aggregate the worker duration and cpu_s fields.
duration refers to the total job execution duration, in seconds
cpu_s is derived from the Process::CLOCK_THREAD_CPUTIME_ID counter, and is a measure of time spent by the job on-CPU.
Divide cpu_s by duration to get the percentage time spend on-CPU.
If this ratio exceeds 33%, the worker is considered CPU-bound and should be annotated as such.
Note that these values should not be used over small sample sizes, but rather over fairly large aggregates.

Feature category

All Sidekiq workers must define a known feature category.

Job weights

Some jobs have a weight declared. This is only used when running Sidekiq in the default execution mode - using sidekiq-cluster does not account for weights.

As we are moving towards using sidekiq-cluster in Core, newly-added workers do not need to have weights specified. They can simply use the default weight, which is 1.

Worker context

Introduced in GitLab 12.8.

To have some more information about workers in the logs, we add metadata to the jobs in the form of an ApplicationContext. In most cases, when scheduling a job from a request, this context is already deducted from the request and added to the scheduled job.

When a job runs, the context that was active when it was scheduled is restored. This causes the context to be propagated to any job scheduled from within the running job.

All this means that in most cases, to add context to jobs, we don't need to do anything.

There are however some instances when there would be no context present when the job is scheduled, or the context that is present is likely to be incorrect. For these instances, we've added Rubocop rules to draw attention and avoid incorrect metadata in our logs.

As with most our cops, there are perfectly valid reasons for disabling them. In this case it could be that the context from the request is correct. Or maybe you've specified a context already in a way that isn't picked up by the cops. In any case, leave a code comment pointing to which context to use when disabling the cops.

When you do provide objects to the context, make sure that the route for namespaces and projects is pre-loaded. This can be done by using the .with_route scope defined on all Routables.

Cron workers

The context is automatically cleared for workers in the Cronjob queue (include CronjobQueue), even when scheduling them from requests. We do this to avoid incorrect metadata when other jobs are scheduled from the cron worker.

Cron workers themselves run instance wide, so they aren't scoped to users, namespaces, projects, or other resources that should be added to the context.

However, they often schedule other jobs that do require context.

That is why there needs to be an indication of context somewhere in the worker. This can be done by using one of the following methods somewhere within the worker:

Wrap the code that schedules jobs in the with_context helper:

  def perform
    deletion_cutoff = Gitlab::CurrentSettings
                        .deletion_adjourned_period.days.ago.to_date
    projects = Project.with_route.with_namespace
                 .aimed_for_deletion(deletion_cutoff)

    projects.find_each(batch_size: 100).with_index do |project, index|
      delay = index * INTERVAL

      with_context(project: project) do
        AdjournedProjectDeletionWorker.perform_in(delay, project.id)
      end
    end
  end

Use the a batch scheduling method that provides context:

  def schedule_projects_in_batch(projects)
    ProjectImportScheduleWorker.bulk_perform_async_with_contexts(
      projects,
      arguments_proc: -> (project) { project.id },
      context_proc: -> (project) { { project: project } }
    )
  end

Or, when scheduling with delays:

  diffs.each_batch(of: BATCH_SIZE) do |diffs, index|
    DeleteDiffFilesWorker
      .bulk_perform_in_with_contexts(index *  5.minutes,
                                     diffs,
                                     arguments_proc: -> (diff) { diff.id },
                                     context_proc: -> (diff) { { project: diff.merge_request.target_project } })
  end

Jobs scheduled in bulk

Often, when scheduling jobs in bulk, these jobs should have a separate context rather than the overarching context.

If that is the case, bulk_perform_async can be replaced by the bulk_perform_async_with_context helper, and instead of bulk_perform_in use bulk_perform_in_with_context.

For example:

    ProjectImportScheduleWorker.bulk_perform_async_with_contexts(
      projects,
      arguments_proc: -> (project) { project.id },
      context_proc: -> (project) { { project: project } }
    )

Each object from the enumerable in the first argument is yielded into 2 blocks:

The arguments_proc which needs to return the list of arguments the job needs to be scheduled with.
The context_proc which needs to return a hash with the context information for the job.

Arguments logging

As of GitLab 13.6, Sidekiq job arguments are logged by default, unless SIDEKIQ_LOG_ARGUMENTS is disabled.

By default, the only arguments logged are numeric arguments, because arguments of other types could contain sensitive information. To override this, use loggable_arguments inside a worker with the indexes of the arguments to be logged. (Numeric arguments do not need to be specified here.)

For example:

class MyWorker
  include ApplicationWorker

  loggable_arguments 1, 3

  # object_id will be logged as it's numeric
  # string_a will be logged due to the loggable_arguments call
  # string_b will be filtered from logs
  # string_c will be logged due to the loggable_arguments call
  def perform(object_id, string_a, string_b, string_c)
  end
end

Tests

Each Sidekiq worker must be tested using RSpec, just like any other class. These tests should be placed in spec/workers.

Sidekiq Compatibility across Updates

Keep in mind that the arguments for a Sidekiq job are stored in a queue while it is scheduled for execution. During a online update, this could lead to several possible situations:

An older version of the application publishes a job, which is executed by an upgraded Sidekiq node.
A job is queued before an upgrade, but executed after an upgrade.
A job is queued by a node running the newer version of the application, but executed on a node running an older version of the application.

Changing the arguments for a worker

Jobs need to be backward and forward compatible between consecutive versions of the application. Adding or removing an argument may cause problems during deployment before all Rails and Sidekiq nodes have the updated code.

Deprecate and remove an argument

Before you remove arguments from the perform_async and perform methods., deprecate them. The following example deprecates and then removes arg2 from the perform_async method:

Provide a default value (usually nil) and use a comment to mark the argument as deprecated in the coming minor release. (Release M)

class ExampleWorker
  # Keep arg2 parameter for backwards compatibility.
  def perform(object_id, arg1, arg2 = nil)
    # ...
  end
end

One minor release later, stop using the argument in perform_async. (Release M+1)
```
ExampleWorker.perform_async(object_id, arg1)
```
At the next major release, remove the value from the worker class. (Next major release)
```
class ExampleWorker
  def perform(object_id, arg1)
    # ...
  end
end
```

Add an argument

There are two options for safely adding new arguments to Sidekiq workers:

Set up a multi-step deployment in which the new argument is first added to the worker.
Use a parameter hash for additional arguments. This is perhaps the most flexible option.

Multi-step deployment

This approach requires multiple releases.

Add the argument to the worker with a default value (Release M).

class ExampleWorker
  def perform(object_id, new_arg = nil)
    # ...
  end
end

Add the new argument to all the invocations of the worker (Release M+1).
```
ExampleWorker.perform_async(object_id, new_arg)
```

Remove the default value (Release M+2).

class ExampleWorker
  def perform(object_id, new_arg)
    # ...
  end
end

Parameter hash

This approach doesn't require multiple releases if an existing worker already uses a parameter hash.

Use a parameter hash in the worker to allow future flexibility.

class ExampleWorker
  def perform(object_id, params = {})
    # ...
  end
end

Removing workers

Try to avoid removing workers and their queues in minor and patch releases.

During online update instance can have pending jobs and removing the queue can lead to those jobs being stuck forever. If you can't write migration for those Sidekiq jobs, please consider removing the worker in a major release only.

Renaming queues

For the same reasons that removing workers is dangerous, care should be taken when renaming queues.

When renaming queues, use the sidekiq_queue_migrate helper migration method, as shown in this example:

class MigrateTheRenamedSidekiqQueue < ActiveRecord::Migration[5.0]
  include Gitlab::Database::MigrationHelpers

  DOWNTIME = false

  def up
    sidekiq_queue_migrate 'old_queue_name', to: 'new_queue_name'
  end

  def down
    sidekiq_queue_migrate 'new_queue_name', to: 'old_queue_name'
  end
end