Commit graph

26 commits

Author SHA1 Message Date
Yorick Peterse
9a20ff1c99
Added Gitlab::Database.config
This returns the ActiveRecord configuration for the current environment.

While CE doesn't use this very often, EE will use it in a few places for
the database load balancing code. I'm adding this to CE so we don't end
up with merge conflicts in this file.
2017-03-17 17:08:10 +01:00
Yorick Peterse
fb697e69f2
Clear AR connections before starting Sidekiq
This should ensure that connections obtained before starting Sidekiq are
not leaked, leading to connection timeouts.

Fixes gitlab-com/infrastructure#1139
2017-03-08 00:25:59 +01:00
Yorick Peterse
f116f87c91
Don't use backup AR connections for Sidekiq
Adding two extra connections does nothing other than increasing the
number of idle database connections. Given Sidekiq uses N threads it can
never use more than N AR connections at a time, thus we don't need more.

The initializer mentioned the Sidekiq upgrade guide stating this was
required.  This is false, the Sidekiq upgrade guide states this is
necessary for Redis and not ActiveRecord.

On GitLab.com this resulted in a reduction of about 80-100 PostgreSQL
connections.

Fixes #27713
2017-02-06 16:35:41 +01:00
Yorick Peterse
88e627cf14
Fix race conditions for AuthorizedProjectsWorker
There were two cases that could be problematic:

1. Because sometimes AuthorizedProjectsWorker would be scheduled in a
   transaction it was possible for a job to run/complete before a
   COMMIT; resulting in it either producing an error, or producing no
   new data.

2. When scheduling jobs the code would not wait until completion. This
   could lead to a user creating a project and then immediately trying
   to push to it. Usually this will work fine, but given enough load it
   might take a few seconds before a user has access.

The first one is problematic, the second one is mostly just annoying
(but annoying enough to warrant a solution).

This commit changes two things to deal with this:

1. Sidekiq scheduling now takes places after a COMMIT, this is ensured
   by scheduling using Rails' after_commit hook instead of doing so in
   an arbitrary method.

2. When scheduling jobs the calling thread now waits for all jobs to
   complete.

Solution 2 requires tracking of job completions. Sidekiq provides a way
to find a job by its ID, but this involves scanning over the entire
queue; something that is very in-efficient for large queues. As such a
more efficient solution is necessary. There are two main Gems that can
do this in a more efficient manner:

* sidekiq-status
* sidekiq_status

No, this is not a joke. Both Gems do a similar thing (but slightly
different), and the only difference in their name is a dash vs an
underscore. Both Gems however provide far more than just checking if a
job has been completed, and both have their problems. sidekiq-status
does not appear to be actively maintained, with the last release being
in 2015. It also has some issues during testing as API calls are not
stubbed in any way. sidekiq_status on the other hand does not appear to
be very popular, and introduces a similar amount of code.

Because of this I opted to write a simple home grown solution. After
all, all we need is storing a job ID somewhere so we can efficiently
look it up; we don't need extra web UIs (as provided by sidekiq-status)
or complex APIs to update progress, etc.

This is where Gitlab::SidekiqStatus comes in handy. This namespace
contains some code used for tracking, removing, and looking up job IDs;
all without having to scan over an entire queue. Data is removed
explicitly, but also expires automatically just in case.

Using this API we can now schedule jobs in a fork-join like manner: we
schedule the jobs in Sidekiq, process them in parallel, then wait for
completion. By using Sidekiq we can leverage all the benefits such as
being able to scale across multiple cores and hosts, retrying failed
jobs, etc.

The one downside is that we need to make sure we can deal with
unexpected increases in job processing timings. To deal with this the
class Gitlab::JobWaiter (used for waiting for jobs to complete) will
only wait a number of seconds (30 by default). Once this timeout is
reached it will simply return.

For GitLab.com almost all AuthorizedProjectWorker jobs complete in
seconds, only very rarely do we spike to job timings of around a minute.
These in turn seem to be the result of external factors (e.g. deploys),
in which case a user is most likely not able to use the system anyway.

In short, this new solution should ensure that jobs are processed
properly and that in almost all cases a user has access to their
resources whenever they need to have access.
2017-01-25 13:22:15 +01:00
Rydkin Maxim
170efaaba2 Enable Style/MultilineOperationIndentation in Rubocop, fixes #25741 2016-12-16 21:37:22 +03:00
Stan Hu
fe0de99c4e Gracefully recover from Redis connection failures in Sidekiq initializer
* Closes gitlab-org/gitlab-ce#25143
* Closes gitlab-org/omnibus-gitlab#1743
2016-12-01 06:24:59 -08:00
Patricio Cano
208530494e Refactored initializer code to its own class and added tests 2016-11-10 11:38:11 -06:00
Patricio Cano
b95216aaba Allow the Sidekiq queues to throttle and the factor by which to throttle them to be configurable 2016-11-10 11:38:11 -06:00
Patricio Cano
9e2964c15a Allow certain Sidekiq jobs to be throttled 2016-11-10 11:38:10 -06:00
Drew Blessing
d498ec98e0 Set default Sidekiq retries to 3
By default, Sidekiq will retry 25 times with an exponential backoff.
This may result in jobs retrying for up to 21 days. Most Sidekiq
failures occur when attempting to connect to external services -
Project service hooks, web hooks, mailers, mirror updates, etc.
We should set a default retry of 3, and if that's not sufficient
individual workers can override this in the worker class.
2016-11-04 14:58:02 -05:00
Stan Hu
cde3963dae Initialize Sidekiq with the list of queues used by GitLab
The Sidekiq client API adds an entry to the Sidekiq "queues" list,
but mail_room and gitlab-shell use redis-rb directly to insert jobs
into Redis and thus do not make an extra "sadd" call to Redis
each time a job is inserted. To make it possible to monitor
these queues via the API, add an initialization step to
set up the list at startup.

Closes gitlab-com/infrastructure#682
2016-11-01 07:03:25 -07:00
Gabriel Mazetto
2625a4ccbd Make sidekiq get config settings from Gitlab::RedisConfig 2016-08-04 18:55:37 +02:00
Stan Hu
22386040fe Enable SIDEKIQ_REQUEST_STORE by default 2016-07-25 18:05:14 -07:00
Stan Hu
f4aac77389 Add support for using RequestStore within Sidekiq tasks via SIDEKIQ_REQUEST_STORE env variable
This significantly reduces the DB churn in the PostReceive task when it
performs reference extraction.

See #18663
2016-07-25 17:59:09 -07:00
Gabriel Mazetto
d7ab81cf7d Log cron_jobs configuration instead of raising exception 2016-07-20 12:50:42 +02:00
Gabriel Mazetto
aaba9ebbb1 Improve cron_jobs loading error messages 2016-07-18 23:52:58 +02:00
Grzegorz Bizon
28bafd5354 Enable Style/SpaceAfterComma Rubocop cop 2016-06-29 15:23:44 +02:00
Rémy Coutable
3cb1e96ef6
Avoid autoload issue such as 'Mail::Parsers::AddressStruct'
By eager-loading the Mail gem in the Sidekiq initializer.

Signed-off-by: Rémy Coutable <remy@rymai.me>
2016-06-20 12:43:29 +02:00
Valery Sizov
2244aaf98f Redis configuration consistency 2016-04-13 20:57:19 +03:00
Jacob Vosmaer
0163e27631 Add Gitlab::Redis connection pool 2016-04-04 17:00:09 +02:00
Jacob Vosmaer
c46e6eff10 Parse config/resque.yml in one place only 2016-03-09 15:16:46 +01:00
Gabriel Mazetto
a7a52f7567 Hotfix for sidekiq-cron being loaded from Settingslogic with defaults 2015-12-22 13:02:42 -02:00
Gabriel Mazetto
51ef4a2fed Sidekiq-cron configuration moved to gitlab.yml 2015-12-14 14:17:20 -02:00
Gabriel Mazetto
3e5b24d46a Upgraded Sidekiq to 4.x 2015-12-10 17:18:03 -02:00
Gabriel Mazetto
496870ddec Migrate from Sidetiq to Sidekiq-cron
Updated Sidekiq to 3.5.x
2015-12-04 11:29:45 -02:00
Douwe Maan
046b283127 Groundwork for merging CI into CE 2015-08-25 18:42:46 -07:00
Renamed from config/initializers/4_sidekiq.rb (Browse further)