Whenever you push to a branch GitLab will show a button to create a
merge request (should one not exist already). The underlying code to
display this data was quite inefficient. For example, it involved
multiple slow queries just to figure out what the most recent push event
was.
This commit changes the way this data is retrieved so it's much faster.
This is achieved by caching the ID of the last push event on every push,
which is then retrieved when loading certain pages. Database queries are
only executed if necessary and the cached data is removed automatically
once a merge request has been created, or 2 hours after being stored.
A trade-off of this approach is that we _only_ track the last event.
Previously if you were to push to branch A and B then create a merge
request for branch B we'd still show the widget for branch A. As of this
commit this is no longer the case, instead we will only show the widget
for the branch you pushed to most recently. Once a merge request exists
the widget is no longer displayed. Alternative solutions are either too
complex and/or too slow, hence the decision was made to settle for this
trade-off.
Performance Impact
------------------
In the best case scenario (= a user didn't push anything for more than 2
hours) we perform a single Redis GET per page. Should there be cached
data we will run a single (and lightweight) SQL query to get the
event data from the database. If a merge request already exists we will
run an additional DEL to remove the cache key.
The difference in response timings can vary a bit per project. On
GitLab.com the 99th percentile of time spent in User#recent_push hovers
between 100 milliseconds and 1 second, while the mean hovers around 50
milliseconds. With the changes in this MR the expected time spent in
User#recent_push is expected to be reduced down to just a few
milliseconds.
Fixes https://gitlab.com/gitlab-org/gitlab-ce/issues/35990
In various places we check if the same relation would return projects.
This is done using "any?" which will run a COUNT query with any
LIMIT/OFFSET values still applied.
To work around all this we introduce 2 helper methods that take care of
doing the right thing. This leads to the produced queries being simpler
and fewer queries being executed.
When sign-in is disabled:
- skip password expiration checks
- prevent password reset requests
- don’t show Password tab in User Settings
- don’t allow login with username/password for Git over HTTP requests
- render 404 on requests to Profiles::PasswordsController
If internal auth is disabled and user is not an LDAP user, present
the user with an alert to create a personal access token if he does
not have one already.
Projects::ForkService delegates to this service almost entirely, but needed
one small change so it would propagate create errors correctly.
CreateService#execute needs significant refactoring; it is now right at the
complexity limit set by Rubocop. I avoided doing so in this commit to keep the
diff as small as possible.
Several tests depend on the insecure behaviour of ForkService, so fi them up at
the same time.
+ Move 'Edit Project/Group' out of membership-related partial
+ Show the access request buttons only to logged-in users
+ Put the request access buttons out of in a more visible button
+ Improve the copy in the #remove_member_message helper
Signed-off-by: Rémy Coutable <remy@rymai.me>
Steps to reproduce:
1. Create a project with a README
2. In the actual remote, type: `git symbolic-ref HEAD refs/heads/nowhere`
3. Check that HEAD is gone via `git ls-remote .`
4. Go to the projects page and see the Error 500
Closes https://github.com/gitlabhq/gitlabhq/issues/9484
The previous behavior was, that the first element of the select was
preselected, thus upon saving the project, the previous selected could
be overwritten.