fix changelog MR reference

add non-HEAD builds finder and add `created` pipelines to scope

add spec for auto-cancel non-HEAD pipelines and refactor create_pipeline_service_spec

more refactoring for spec

adds option for auto-cancel into CI/CD settings

fix spec to new configuration

fix rubocop

fix schema.rb

fix schema.rb

replace Gitlab 9.0 with 9.1 in doc

change wording on pipeline settings

added auto_canceled_by field as identifier of autocancel subject

remove unnecessary index

replace service with retry_lock

replace auto_cancel_pending_pipelines boolean setting with integer (and enum in model)

fix schema.rb

fix schema.rb

remove projekt attribute and clean up spec

clean up spec withcouple of shared examples

added spec for "It does not cancel current pipeline" scenario

add some specs to auto-cancel

add spec for another branch pipelines

Signed-off-by: NRémy Coutable <remy@rymai.me>

Signed-off-by: NRémy Coutable <remy@rymai.me>

Signed-off-by: NRémy Coutable <remy@rymai.me>

Signed-off-by: NRémy Coutable <remy@rymai.me>

Implement module for setting "Poll-Interval" response header.
Return 429 in ETag caching middleware when polling is disabled.

Signed-off-by: NRémy Coutable <remy@rymai.me>

This is really weird. I think we did get the same issue while
migrating on GitLab.com. We fixed this by adding job_id to
ci_builds table manually, and then run the migrations again.
However I think we didn't hit into this on staging, which should
somehow be a production clone.

At any rate, I guess we could check if the column exists in the
migration. If the column is not there, there's no point to remove
that column anyway.

Closes #29976

For example:

    index_exists?(:routes, name: INDEX_NAME) # => false
    index_exists?(:routes, :path, name: INDEX_NAME) # => true

Nested groups support uses queries along the lines of `path LIKE 'X/%'`.
For these queries to use an index on PostgreSQL we need to use either
the varchar_pattern_ops or text_pattern_ops operator class.

Fixes https://gitlab.com/gitlab-org/gitlab-ce/issues/29554

Signed-off-by: NDmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

Signed-off-by: NRémy Coutable <remy@rymai.me>

This reverts commit 68e40bd4, reversing
changes made to 2d1f823b.

This reverts commit 5e966688, reversing
changes made to b35378a9.

when rolling back from 9-0-stable to 8.17.3

Feedback:
https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/9908#note_25324225

mysql cannot simply drop the index without dropping the corresponding
foreign key, and we certainly don't want to drop the foreign key
here.

Closes #29106

That have wildcard routes associated and not reserved yet: artifacts,
graphs, badges and refs
Signed-off-by: NDmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

This migration resets `users.authorized_projects_populated` (in batches)
so that active users' permissions are refreshed on the next request.
This ensures that their permissions are properly stored.

Given GitLab moves away from using the term build, everywhere, also the
CI variables are being renamed.

For now, both `CI_BUILD_X` and `CI_JOB_X` are supported, with the same
values. However, in about 3 months, support will be dropped.