On .com repositories are created through Gitaly for a while now. For
customers this is not the case unless these have chosen to do so through
feature flags. By moving this to opt out, everyone will be using this.
This move is part of the migration issue
https://gitlab.com/gitlab-org/gitaly/issues/593
The bigger impact this commit will have is that tests that use a
repository through `FactoryBot.create(:project, :repository)` will now
use Gitaly to do so. As tests run on the same disk, or at least machine,
this will most probably slow them down.
Prior to this change, this method was called add_namespace, which broke
the CRUD convention and made it harder to grep for what I was looking
for. Given the change was a find and replace kind of fix, this was
changed without opening an issue and on another feature branch.
If more dynamic calls are made to add_namespace, these could've been
missed which might lead to incorrect bahaviour. However, going through
the commit log it seems thats not the case.
By default, --prune is added to the command-line of a `git fetch` operation,
but for repositories with many references this can take a long time to run. We
shouldn't need to run --prune the first time we fetch a new repository.
Migration is done through a small refactoring, which makes us call
endpoins which are performing the same actions for namespaces.
Tests are added to ensure only the project is removed that should be
removed.
Closesgitlab-org/gitaly#873
Originally from branch 'fix-authorized-keys-enabled-default-2738' via merge request https://gitlab.com/gitlab-org/gitlab-ee/merge_requests/2240
Removed background migrations which were intended to fix state after using Gitlab
without a default having been set
Squashed commits:
Locally, if Spring was not restarted, `current_application_settings` was still cached, which prevented the migration from editing the file. This will also ensure that any app server somehow hitting old cache data will properly default this setting regardless.
Retroactively fix migration
This allows us to identify customers who ran the broken migration. Their `authorized_keys_enabled` column does not have a default at this point.
We will fix the column after we fix the `authorized_keys` file.
Fix authorized_keys file if needed
Add default to authorized_keys_enabled setting
Reminder: The original migration was fixed retroactively a few commits ago, so people who did not ever run GitLab 9.3.0 already have a column that defaults to true and disallows nulls. I have tested on PostgreSQL and MySQL that it is safe to run this migration regardless.
Affected customers who did run 9.3.0 are the ones who need this migration to fix the authorized_keys_enabled column.
The reason for the retroactive fix plus this migration is that it allows us to run a migration in between to fix the authorized_keys file only for those who ran 9.3.0.
Tweaks to address feedback
Extract work into background migration
Move batch-add-logic to background migration
Do the work synchronously to avoid multiple workers attempting to add batches of keys at the same time.
Also, make the delete portion wait until after adding is done.
Do read and delete work in background migration
Fix Rubocop offenses
Add changelog entry
Inform the user of actions taken or not taken
Prevent unnecessary `select`s and `remove_key`s
Add logs for action taken
Fix optimization
Reuse `Gitlab::ShellAdapter`
Guarantee the earliest key
Fix migration spec for MySQL
Originally branch 'mk-toggle-writing-to-auth-keys-1631'
See merge request !2004
Squashed commits:
Add authorized_keys_enabled to Application Settings
Ensure default settings are exposed in UI
Without this change, `authorized_keys_enabled` is unchecked when it is nil, even if it should be checked by default.
Add “Speed up SSH operations” documentation
Clarify the reasons for disabling writes
Add "How to go back" section
Tweak copy
Update Application Setting screenshot
By importing this Ruby code into gitlab-rails (and gitaly-ruby), we avoid
200ms of startup time for each gitlab_projects subprocess we are eliminating.
By not having a gitlab_projects subprocess between gitlab-rails / sidekiq and
any git subprocesses (e.g. for fork_project, fetch_remote, etc, calls), we can
also manage these git processes more cleanly, and avoid sending SIGKILL to them
GitLab uses the import_project method in GitLab Shell,
This method uses a timeout for the operation, hardcoded to 800 seconds.
With this MR the timeout is now configurable in the gitlab_shell
settings.
Notice what happens when a user adds a key with a space in between:
```
irb(main):004:0> 'ssh-rsa foobar'.split(/ /)
=> ["ssh-rsa", "", "foobar"]
```
This would cause gitlab-shell to receive a blank argument for the actual
key, leading to users unable to login.
Jacob Vosmaer
Zeger-Jan van de Weg
Stan Hu
Sean McGivern
Douwe Maan
Sean McGivern
Michael Kozono
Alejandro Rodríguez
Nick Thomas
Jacopo
Kim "BKC" Carlbäcker
Grzegorz Bizon
Chris Wilson
Robert Speicher
Douglas Barbosa Alexandre
Robin Bobbitt
Douwe Maan
Rémy Coutable
James Lopez
Dmitriy Zaporozhets