This fix ensures tags that are associated with a release (whether they
have a backslash in their name or not) will link to that single release
page instead of the releases index page, since releases are paginated
and therefore any older releases will often have a broken link as they
end up on a different page.

Eduardo Sanz-Garcia authored 2 years ago and Marc Shaw committed 2 years ago

Because of the possible risk introduce in this MR, we are going to
conditional turn on/off this feature using the flag: `access_token_ajax`

See: https://gitlab.com/gitlab-org/gitlab/-/issues/359956
Complete plan: https://gitlab.com/gitlab-org/gitlab/-/issues/357848

Manoj M J authored 2 years ago and Bob Van Landuyt :neckbeard: committed 2 years ago

Introduce new method to refresh authorizations
on group member

Extract ServicePing payload computation and persistance
out of SubmitService into GitlabServicePingWorker
in order to pre compute SerivcePing data for preview
for instance admins even if those instances does not submit
this data later on to GitLab

Changelog: performance

Uses the application settings value stored into
the `pipeline_limit_per_project_user_sha` column
to control the pipelines creation rate limit.

To enable the rate limit the value needs to be
greater than zero and the feature flag
`ci_enforce_throttle_pipelines_creation` has to be enabled.
Otherwise the rate limit will only log the pipelines that
are over the limit without preventing them from being created.

Changelog: other

Remove redundant namespaces from a base query when searching for
descendants.

João Pereira authored 2 years ago and David Fernandez committed 2 years ago

Adds a new container_registry_migration_phase2_capacity_40
feature flag to enable a capacity of 40 concurrent migrations.

Contributes to https://gitlab.com/gitlab-org/gitlab/-/issues/362263

**Problem**

The worker executes many read SQL queries to the primary database. We
prefer to send them to replica instead.

**Solution**

Use `sticky` data consistency for the worker

- Address N+1s
- Modify test to use authorisations table properly
- Add faster_owner_access feature flag

alex pooley authored 2 years ago and Luke Duncalfe committed 2 years ago

https://gitlab.com/gitlab-org/gitlab/-/issues/321947
https://gitlab.com/gitlab-org/gitlab/-/issues/323312

Changelog: added

Adds initial support for configuring internal IP ranges in
addition to the custom IP lists at group level. For now this
is added as an environment variable but in future would likely
be an application setting to configure the in-use internal
network IP ranges.

Changelog: fixed

Katrin Leinweber authored 2 years ago and Fabio Pitino committed 2 years ago

Sorting projects with most CI minutes to the top of the list,
seems more sensible for a "quota" tab than sorting by project ID.
This may help customers find possibly inefficient CI scripts.

Co-authored-by: Firdaws Farukh <ffarukh@gitlab.com>

Changelog: changed
EE: true

When consolidated storage or IAM instance profiles are in use for S3,
Workhorse uses its own S3 client and handles multipart uploads without
needing the pre-generated values from Rails. The problem is that these
pre-generated multipart URLs were never completed or aborted if the S3
client were in use, leaving a trail of many unterminated uploads.

This feature flag is enabled by default since this change only affects
AWS S3 users. This should reduce the number of incomplete multipart
uploads created by GitLab.

We previously tried to omit the multipart URLs in
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/49482, but had
to revert this in
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/52561 since
Workhorse would only use this client if the region were
specified. However, with
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/59932 a default
region will always be sent to Workhorse.

Relates to https://gitlab.com/gitlab-org/gitlab/-/issues/342136

- ci_expose_running_job_token_for_artifacts
- ci_authenticate_running_job_token_for_artifacts

Changelog: changed

MR: gitlab.com/gitlab-org/gitlab/-/merge_requests/86612
Issue: gitlab.com/gitlab-org/gitlab/-/issues/358237

This will use a git clone for the first sync for a
repository instead of git fetch.

Changelog: other
EE: true

This change will default enable the feature flag
`automated_email_provision`. Behind this feature flag are two banners
for Offline Cloud Licensing that will be displayed in the Admin area
on the dashboard (`/admin`) and on the subscription page
(`/admin/subscription`).

Changelog: added
EE: true

With 4aba970e (gitaly_client: Allow filtering quarantined commits by
existence, 2022-05-03), we have introduced code that enables us to use
the repo's quarantine directory in order to enumerate commits in access
checks. This can lead to a significant speedup when enumerating commits
in repositories which have a lot of references because the previous way
of enumerating commits would require us to walk each of the preexisting
refs so that we can find the commits that really are new.

We have rolled the accompanying feature flag without any observed issues
so far, so let's default-enable the feature flag.

Changelog: changed

Changelog: added

Eugenia Grieff authored 2 years ago and Charlie Ablett committed 2 years ago

Update docs

Changelog: added

Instead of the sidebar on merge requests being globally attached
to the right of the screen, this changes that by moving it
into the overview tab.

Contributes to https://gitlab.com/gitlab-org/gitlab/-/issues/225174

* Add BlameService to control pagination and blame range calculations
* Add Kaminari pagination for File Blame

Fixes some alignment issues when on mobile

This change builds the deploy web hook payload in the request, rather
than delayed through `Deployments::HooksWorker`.

Once the payload is built, it is sent to the queue via `WebHookWorker`
and the webhook executes on the queue, as normal.

This change is behind a feature flag to allow us to test whether the
change will fix a problem for a particular customer who has some webhook
payloads containing a "status" of "running" when the deployment has
finished, when the status should be "succeeded".

https://gitlab.com/gitlab-org/gitlab/-/issues/355903

Sean Arnold authored 2 years ago and Mayra Cabrera committed 2 years ago

Changelog: removed

Eduardo Bonet authored 2 years ago and Mayra Cabrera committed 2 years ago

Tracks notes left on Ipynb notes either
for commits or mrs

This sets the data_consistency for the Ci::ArchiveTraceWorker to
sticky. Allowing the read queries before any write to be handled by a
replica.

The job is scheduled from the Ci::BuildFinishedWorker with some delay
built in. Which means this job can take the 0.8s wait in the
beginning, and is very likely to find a replica that is up to date
with the WAL locations at the time of scheduling.

To increase the effect of this, this also preloads some associations
in the worker. That way, we can take advantage of loading them from a
replica rather than doing it lazily from the primary.

David Fernandez authored 2 years ago and Bob Van Landuyt :neckbeard: committed 2 years ago

In loop mode, the Enqueuer will loop on migration starts until either:
* it fills the capacity
* it hits the loop deadline
* the loop mode feature flag is disabled

Furkan Ayhan authored 2 years ago and Grzegorz Bizon committed 2 years ago

When the right side of a comparison is a regexp variable, we consider
this as a string. However, we need to try converting it to a regexp
if possible.

This makes sure that the ProjectImportScheduleWorker uses replicas for
it's reads in the beginning of the job.

We need to allow retries for the `delayed` strategy, because that will
raise an error if it cannot find an up-to-date replica in 0.8s the
first time.

If the job can't find an up-to-date replica in the retry, it will use
the primary instead.

Abdul Wadood authored 2 years ago and Adam Hegyi committed 2 years ago

Async indexes are required because the events table is too big.
Partial indexes are used to reduce the size of the resulting indexes.

Changelog: performance

This commit cleans up ci_reduce_persistent_ref_writes
feature flag.

Changelog: other

Lee Tickett authored 2 years ago and Charlie Ablett committed 2 years ago

Changelog: added

This change will hide certificate-based cluster connection related
features by default globaly. The FF can still be used to bring the
features back if needed. This is part of the effor of
https://gitlab.com/groups/gitlab-org/-/epics/6940

Changelog: changed

Removes the Monitor Logs feature by moving all
related code behind a feature flag.

Changelog: removed