gitlab-org--gitlab-foss/config/sidekiq_queues.yml
Toon Claes 0ad80cab40 Use worker to destroy namespaceless projects in post-deploy
Destroying projects can be very time consuming. So instead of destroying them in
the post-deploy, just schedule them and make Sidekiq do the hard work.

They are scheduled in batches of 5000 records. This way the number of database
requests is limited while also the amount data read to memory is limited.
2017-05-10 15:01:27 +02:00

57 lines
1.6 KiB
YAML

# This configuration file should be exclusively used to set queue settings for
# Sidekiq. Any other setting should be specified using the Sidekiq CLI or the
# Sidekiq Ruby API (see config/initializers/sidekiq.rb).
---
# All the queues to process and their weights. Every queue _must_ have a weight
# defined.
#
# The available weights are as follows
#
# 1: low priority
# 2: medium priority
# 3: high priority
# 5: _super_ high priority, this should only be used for _very_ important queues
#
# As per http://stackoverflow.com/a/21241357/290102 the formula for calculating
# the likelihood of a job being popped off a queue (given all queues have work
# to perform) is:
#
# chance = (queue weight / total weight of all queues) * 100
:queues:
- [post_receive, 5]
- [merge, 5]
- [update_merge_requests, 3]
- [process_commit, 3]
- [new_note, 2]
- [build, 2]
- [pipeline, 2]
- [gitlab_shell, 2]
- [email_receiver, 2]
- [emails_on_push, 2]
- [mailers, 2]
- [upload_checksum, 1]
- [use_key, 1]
- [repository_fork, 1]
- [repository_import, 1]
- [project_service, 1]
- [delete_user, 1]
- [delete_merged_branches, 1]
- [authorized_projects, 1]
- [expire_build_instance_artifacts, 1]
- [group_destroy, 1]
- [irker, 1]
- [namespaceless_project_destroy, 1]
- [project_cache, 1]
- [project_destroy, 1]
- [project_export, 1]
- [project_web_hook, 1]
- [repository_check, 1]
- [system_hook, 1]
- [git_garbage_collect, 1]
- [reactive_caching, 1]
- [cronjob, 1]
- [default, 1]
- [pages, 1]
- [system_hook_push, 1]
- [update_user_activity, 1]
- [propagate_service_template, 1]